# What is GEO? Canonical: https://whatisgeo.app/guide/ Publisher: Adapt Progress Evolve Limited — https://adaptprogressevolve.com/ Contact: hello@whatisgeo.app Updated: 14 September 2026 ## What is GEO? Generative Engine Optimisation (GEO) is the practice of improving how visible and accurately represented your website or content is in answers generated by AI systems like ChatGPT, Google's AI Overviews, or Perplexity. That includes being mentioned, being cited, and sometimes getting a referral click. None of those outcomes are guaranteed. GEO influences the odds; it doesn't fix them. GEO isn't a new mystical discipline that replaces everything you already know about being findable online. It's an extension of it. If your site is already thin, unclear, or badly structured, no amount of GEO tinkering will save it. If it's solid, GEO is about making sure AI systems can find, trust, and quote the good parts. So this isn't a page selling you a shortcut. It's a page explaining the mechanics, the actual research, and what you can do this month that's worth your time. ## GEO and SEO GEO doesn't replace SEO. It sits on top of it, and overlaps heavily with a related idea called AEO (answer engine optimisation). Same foundation, different output format. | | SEO | GEO | |---|---|---| | **Goal** | Rank in search results | Be accurately represented in AI-generated answers | | **Format** | Blue links, snippets, rich results | Synthesised answers, sometimes with citations | | **Signals** | Crawlability, relevance, authority, technical health | Overlapping content and technical foundations; selection varies by system | | **Measurement** | Rankings, impressions, clicks (Search Console) | Manual prompt checks, mentions, citations, referral traffic | Google's own guidance on AI features says existing SEO best practice applies: your page needs to be indexed and eligible for a search snippet to be a supporting link in AI Overviews or AI Mode. That doesn't mean every AI platform uses the same ranking or selection process. There's no special AI schema, no secret file that unlocks inclusion. You can read Google's guide directly at [developers.google.com/search/docs/appearance/ai-features](https://developers.google.com/search/docs/appearance/ai-features). Fair enough, you might think, so why does GEO need its own name at all? Because the selection and synthesis step, how an AI system picks and stitches together passages, behaves differently to how a ranking algorithm sorts ten blue links. That difference is worth understanding on its own terms. That's what the rest of this page is for. ## How AI answers happen Not every AI answer involves a live search. Some come straight from what the model learned during training. Training and live retrieval are different things, and it matters which one is happening when you're thinking about visibility. When a system does search the live web to answer you, it roughly follows four stages. 1. **Query.** The system interprets what you're actually asking. It may rewrite your question or split it into several related searches behind the scenes (sometimes called query fan-out), rather than treating it as one single string. 2. **Discover.** It retrieves a set of candidate sources from the web, using much the same crawling and indexing infrastructure as ordinary search. 3. **Select.** It chooses specific passages from those candidates, favouring content that answers the question directly and reads as reliable. 4. **Answer.** It synthesises a response from the selected passages, sometimes showing citations or links back to sources, sometimes not. This is a simplified model of a search-backed answer, not a published selection formula. Different systems combine retrieval and generation differently. Updating a page can make new information available to search; it doesn't immediately change a model's training. ## The principles ### 1. Answer real customer questions, directly Start with what people genuinely ask, not what you wish they asked. Look at your actual support tickets, sales calls, and reviews for the phrasing customers use. Then answer that question early and plainly, in language a stranger would understand without background context. AI systems are selecting passages that resolve a query cleanly; a page that circles the topic for three paragraphs before answering is a weaker candidate than one that answers in the first two sentences and then elaborates. This isn't about tricking anything. It's about not making the reader, human or machine, hunt for the point. ### 2. Make your content self-contained A passage pulled out of context still needs to make sense. Avoid answers that depend on "as mentioned above" or images with no text equivalent. If a paragraph would be confusing lifted onto someone else's screen with no surrounding page, rewrite it. State who the advice is for, what it assumes, and what it doesn't cover, all inside the same block of text. ### 3. Back claims with real data and explain your method Original data, explained methodology, and specific figures give a page something to cite. Research from Aggarwal et al. (KDD 2024, [arxiv.org/abs/2311.09735](https://arxiv.org/abs/2311.09735)) found that adding citations, quotations, and statistics could improve visibility in benchmark tests, with results varying a lot by domain and topic. That's a research finding under test conditions, not a promise about how any specific production AI system behaves today. Treat it as a reasonable, not guaranteed, lever. ### 4. Show who's behind the words Make it easy for readers to assess who is responsible for the information. Name a real person or a real organisation. Give them a bio, a role, a reason to be believed. This is a practical credibility recommendation, not proof of a universal AI ranking factor. ### 5. Keep facts current and mark corrections When a fact changes (a price, a regulation, a product spec) update the page and date the change. Don't quietly overwrite; note it. Freshness signals matter more for volatile topics than static ones, so judge accordingly. A page about a fixed historical event doesn't need weekly dating. A page about pricing does. ### 6. Fix the plumbing: structure, crawlability, honesty Clean HTML text, semantic headings, a canonical tag, a working sitemap, accurate structured data. None of this is exotic; it's the same technical hygiene good SEO has asked for for years. What's not on this list: keyword stuffing, fabricated citations, bought reviews, or paying anyone for guaranteed inclusion. Those aren't grey-area tactics. They mislead readers and can breach search platform policies. ## Let the right crawlers in Blocking a crawler and welcoming one look almost identical from the outside. Both involve a robots.txt file, a status code, some quiet server-side logic nobody outside IT ever reads. The difference is in the configuration. Check it before assuming your content is accessible. ### Robots.txt is a request, not a lock Robots.txt tells compliant crawlers what they're allowed to fetch. That's it. It's a note on the door, not a bolt. Any crawler that ignores it, or spoofs its user agent string, walks straight past. So if you're relying on robots.txt as your access control layer, you don't have one. That matters for firewall and CDN rules too. Don't allowlist traffic based on the user agent string alone; it's trivial to fake. If you need to verify a crawler is genuinely who it claims to be, check the vendor's published IP ranges or verification method instead. And no, a robots rule won't necessarily cover every user-triggered visit; some providers treat those separately from their standard crawl. ### Search retrieval and model training are different permissions OpenAI documents this split clearly: OAI-SearchBot handles search retrieval, GPTBot handles training data collection, and ChatGPT-User covers visits triggered directly by a user's request inside ChatGPT. These can be managed separately. Allowing OAI-SearchBot to surface your pages in search results does not mean you've agreed to GPTBot training on your content. Full detail sits at [OpenAI's crawler docs](https://developers.openai.com/api/docs/bots). Perplexity works differently again. PerplexityBot exists to surface and link sites within answers, not to train a foundation model. Details are at [Perplexity's crawler docs](https://docs.perplexity.ai/docs/resources/perplexity-crawlers). Two vendors, two different splits. Read the actual docs before assuming your rules do what you think they do. ### llms.txt: useful, not magic llms.txt is a proposed file offering a plain-text summary and navigation aid for AI systems. It's optional. It doesn't replace indexable HTML content, and it doesn't replace your sitemap. Nobody can promise it improves citation or ranking; it isn't an established ranking factor. whatisgeo.app includes one, alongside a full text version kept aligned with the live guide, because it costs little and helps orientation. Fair enough. ### Practical checks worth doing this week Run through these without needing a developer's background, though your developer can help confirm results: 1. Open your main page logged out. Confirm the full answer text loads in the HTML, not hidden behind a script that only fires after interaction. 2. Check the response code. The final intended content page should return 200. Check unexpected redirects and soft 404s. 3. Confirm the canonical tag points where you expect. A wrong canonical can point search engines towards a different preferred page. 4. Use Search Console's URL Inspection tool on your important pages. Site: search isn't comprehensive, so don't trust it as your only signal. 5. Keep your sitemap current. It supports discovery; it doesn't guarantee indexing or inclusion anywhere. None of this reveals a ranking algorithm, and none of it requires bespoke AI schema markup or guarantees priority placement. It's just the groundwork that determines whether the right systems can see your content at all. At the end of the day, that's the whole job: get out of the crawler's way, check your assumptions, and don't confuse a polite request with a locked door. ## A practical example Picture a small furniture shop, sells desks, run by two people and a website. A customer types: "Will this desk fit a small home office?" Generic marketing copy answers that question badly. "Our stylish desk brings functionality to any space" tells the reader nothing they can act on. Here's a rewrite that does better. *(Illustrative values only, not a real product.)* > The Elmswood desk measures 120cm wide, 60cm deep, 75cm tall. Measure your available floor space, chair clearance and access before ordering. The top is 120 × 60cm; compare that with your monitor stands and other equipment. Room area alone won't tell you whether it fits. What changed and why: the answer is self-contained (a passage lifted alone still makes sense), it states a limitation honestly instead of hiding it, and it gives numbers a system can quote directly rather than paraphrase vaguely. None of that guarantees an AI system selects or cites this shop over a competitor's page. But it gives the passage a genuine chance of being useful enough to select, which is the only lever available. Use your real specifications and limitations when applying this to your own pages. ## Your first 30 days A realistic beginner plan. No agency required, no budget assumed. ### Week 1: audit what you already have - List your top 15 customer questions, pulled from real emails, calls, or reviews. - Check whether your existing pages answer any of them directly and self-contained. - Check your important pages with URL Inspection in Google Search Console. A `site:yourdomain.com` search is a quick clue, not a complete indexing audit. What to verify: that Google can actually find and index your key pages at all. If it can't, nothing downstream matters yet. ### Week 2: fix the foundations - Add or tidy semantic headings (H1, H2, H3) on your 5 most important pages. - Set canonical tags where duplicate or near-duplicate pages exist. - Submit or update your XML sitemap. What to verify: that each key page has one clear H1, a logical heading structure beneath it, and no canonical conflicts confusing search engines about which version is the "real" one. ### Week 3: rewrite for direct answers - Pick your 5 highest-value questions from week 1. - Rewrite the top of each relevant page to answer that question in the first two or three sentences. - Add author or business identity information where it's missing (an about page, a named contact, a real address). What to verify: read each rewritten answer out loud, alone, with no other page context. Does it still make sense and answer the question fully? ### Week 4: add substance and start watching - Add one genuinely original piece of data, explanation, or detail per page (a real measurement, a real process, a real limitation), with a note on how you know it. - Pick 3 to 5 prompts real customers might type into an AI tool and run them manually. - Record what comes back: which platform, which date, what was asked, whether you were mentioned, whether you were cited, and in what context. What to verify: nothing dramatic yet. You're establishing a baseline, not expecting results in week four. That's normal. ## How to measure progress A vendor's visibility score describes its own sample of prompts and platforms. It isn't a universal GEO rank. Start with a repeatable manual baseline, then judge whether a paid tool measures what you need. **The core method:** pick a fixed set of prompts that real customers might plausibly ask, and repeat them over time, unchanged. For each one, record: 1. Platform (ChatGPT, Google AI Overview, Perplexity, etc.) 2. Date 3. Exact query text used 4. Whether you were mentioned 5. Whether you were cited or linked 6. The surrounding context (were you framed favourably, neutrally, or not at all) Do this monthly, or after major site changes, using the same prompts each time so you're comparing like with like. Two things to keep in mind. First, answers vary. Ask the same question twice and you may get different phrasing, different sources, sometimes a different answer entirely. Some of this is personalisation; the system may weight your location, history, or account differently to someone else's. A single check is a sample, not a universal ranking, and it isn't reproducible for every user everywhere. Second, separate this from your existing analytics. Search Console reports search impressions, clicks and position, including Google's AI features within its search reporting. That is different from a complete cross-platform citation audit. Track referral traffic from AI platforms in your analytics tool separately when the referring platform passes that information, and treat actual conversions from that traffic as the metric that matters most. A mention with no click and no action is a nice data point, not a business result. | What to track | Where | Frequency | |---|---|---| | Fixed prompt mentions/citations | Manual, across AI platforms | Monthly | | Search rankings & impressions | Search Console | Ongoing, as normal | | AI referral traffic | Analytics referral reports | Monthly | | Referral conversions | Analytics / CRM | Monthly | ## Common questions ### What's the actual difference between GEO and SEO? SEO gets your page ranking in traditional search results. GEO is about being accurately mentioned or cited inside an AI-generated answer. They share the same technical and content foundations; GEO is an additional lens on top, not a separate discipline built from scratch. ### Is SEO dead now that AI answers exist? No. Google's own guidance confirms that a page must be indexed and eligible for a normal search snippet before it can feed AI Overviews or AI Mode. SEO is still the foundation everything else stands on. Anyone telling you otherwise is selling something. ### Do I need to write special code for GEO? No special schema or dedicated AI text file is required. Clean HTML, semantic headings, working canonical tags, and an accurate sitemap are what actually help, and they're the same things good SEO has always asked for. ### What does GEO cost? It depends entirely on what you're starting from. If your content and site structure are already decent, it's mostly time: rewriting for clarity, adding real detail. If you're starting from a broken or thin site, expect more foundational work first. No tool or vendor guarantees results, so spend cautiously and judge by outputs, not promises. ### How long before I see results? There's no fixed timeline, and nobody honest will give you one. Give changes at least a few weeks to be recrawled and reconsidered, then check your fixed prompt set monthly. Treat early months as baseline-setting, not results-chasing. ### Do I need an llms.txt file? It's an optional, experimental proposal (see [llmstxt.org](https://llmstxt.org/)) for offering AI systems a machine-readable summary of your site. It's not a proven ranking factor, it doesn't grant crawl permission on its own, and it doesn't replace robots.txt or your sitemap. Fine to add once your fundamentals are solid; not urgent. ### Can you guarantee we'll be cited by ChatGPT or Google? No, and be wary of anyone who says yes. Citation and mention outcomes vary by platform, query, and time, and no legitimate practice can promise placement in someone else's generated answer. ### What's the difference between AI training and AI search? Training is how a model learned language and facts from a large historical dataset, fixed at a point in time. Live retrieval or search is a separate live process where the system fetches and reads current web pages to answer a specific query. Updating your site may affect future retrieval after it is discovered. It doesn't update a trained model immediately; future training decisions are separate. ### Is any of this realistic for a small business with no dedicated marketing team? Yes. Start with a narrow topic where you have real knowledge. Most of the useful work here is writing honestly and clearly about what you actually offer, which any small business owner can do without specialist tools. ### How often should I update GEO-related content? Whenever a genuine fact changes: prices, specs, availability, regulations. Date the correction. Static, non-volatile pages don't need constant touching just for the sake of activity; that's noise, not signal. --- # Guide: AEO vs SEO: what is the difference? Canonical: https://whatisgeo.app/aeo-vs-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > SEO is the wider work of making a site findable in search. AEO is a label for shaping content so answer engines can quote it. Google says its AI features need no special optimisation beyond good SEO. Treat AEO as an emphasis within SEO. It rests on the same foundations, pages that can be crawled, indexed and understood, and adds attention to how cleanly each page answers a specific question, so a system assembling an answer can quote it or name you. On this page we label three kinds of claim: what a platform documents, what a research study found, and what we recommend. Most pages on this topic blur them together. ## What people mean by AEO AEO stands for answer engine optimisation. There is no formal standard behind it and no specification that any platform maintains. Google's own guide to generative AI search does mention the term, describing AEO and GEO as "terms you may see used to describe work specifically focused on improving visibility in AI search experiences." Google is describing how the words get used; it does not set rules for them. In our observation, usage varies between writers. Some include voice assistants and featured snippets. Some use AEO interchangeably with GEO. Others mean something narrower: content shaped so a chatbot can lift a clean answer from it. If someone offers you "AEO", ask which of these they mean and what exactly they will change on your site. ## What SEO covers SEO is the broader discipline. It covers technical access, so search engines can fetch and index your pages; content that answers real queries; site structure and internal linking; page experience; and the signals that make a source look credible. It predates answer engines by decades, and the core logic holds: be findable, be useful, be credible. We set out the principles we work from in [our GEO principles](/guide/#principles). ## Where AEO and SEO overlap Most of the work. An answer engine cannot quote a page it cannot reach, and a page that is not indexed will not surface in a search feature. Clear writing helps a human reader and a summarising system alike. Consistent facts help everywhere. Google says this directly about its own AI features (documented behaviour). Its Search Central page on AI features states that "The best practices for SEO remain relevant for AI features in Google Search" and that "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." To be eligible as a supporting link, "a page must be indexed and eligible to be shown in Google Search with a snippet." So if your SEO foundations are weak, an AEO layer on top will not rescue them. For the closely related comparison, see [GEO vs SEO](/guide/#geo-vs-seo). ## Where they differ in practice The differences are mostly about emphasis and about who is doing the reading. ### What success looks like Classic SEO usually aims at a click from a ranked result. AEO, as people use the term, aims at being named or quoted inside an answer, which may or may not send anyone to your site. That changes what you measure (see below). ### How you write Question-led headings with the direct answer first suit a skimming reader and a system pulling out a passage. That is our recommendation and nothing more. Google's guide also says "You don't need to write in a specific way just for generative AI search", so think of it as good writing practice that Google does not require. ### Which crawlers you manage AI products use several user agents, each with its own job, and robots.txt can treat each one separately. This is what each company documents: | Company | User agent | What the company says it is for | | --- | --- | --- | | OpenAI | OAI-SearchBot | Surfacing websites in ChatGPT's search features | | OpenAI | GPTBot | Training; disallowing it signals content should not be used to train its foundation models | | OpenAI | ChatGPT-User | Fetches a user asks for; not automatic crawling, and robots.txt rules may not apply | | Perplexity | PerplexityBot | Surfacing and linking websites in Perplexity search results; not used to train foundation models | | Perplexity | Perplexity-User | Fetches for user actions; generally ignores robots.txt because a user requested the fetch | | Anthropic | ClaudeBot | Collecting web content that could contribute to training | | Anthropic | Claude-User | Retrieving pages when people ask Claude questions | | Anthropic | Claude-SearchBot | Indexing to improve search result quality | OpenAI says its settings are independent: a site can allow OAI-SearchBot to appear in search results while disallowing GPTBot. The practical risk is a blanket rule aimed at training crawlers that also shuts out a search crawler you wanted. More on this in [crawler access](/guide/#crawler-access). ## What the platforms themselves say Google's guide to optimising for generative AI features on Google Search is the most direct statement we have found from any platform on this subject. Among other things, it says: - You do not need new machine readable files, "AI text files, markup, or Markdown" to appear in Google Search, and Google Search itself does not use them. - "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." Google still recommends structured data for rich result eligibility, and its AI features page asks that structured data matches the visible text on the page. - "There's no requirement to break your content into tiny pieces for AI to better understand it." - Prioritise "effective SEO strategies" over what it calls "AEO/GEO hacks", and "Be wary of third-party tools that promise ranking success or claim to use 'internal' Google metrics." OpenAI, Perplexity and Anthropic each publish crawler documentation. It covers how their bots identify themselves and how you can allow or restrict them. robots.txt itself is defined in RFC 9309, the Robots Exclusion Protocol. The RFC describes rules that crawlers "are requested to honor" and states that "These rules are not a form of access authorization." So it works only as far as a crawler chooses to follow it. Anthropic says its bots honour robots.txt; OpenAI and Perplexity both say their user-requested fetchers may not apply it. None of this documentation offers a method that guarantees inclusion in an answer. Our stance is that no such guarantee exists, from any platform or any provider. ## What research says, and its limits The most cited study is Aggarwal et al., "GEO: Generative Engine Optimization", accepted at KDD 2024. The authors tested nine ways of rewriting a source: Authoritative, Statistics Addition, Keyword Stuffing, Cite Sources, Quotation Addition, Easy-to-Understand, Fluency Optimization, Unique Words and Technical Terms. They measured how visible each rewritten source became in generated answers, using a word count weighted by position in the answer and a model-scored "subjective impression". The conditions matter. Their benchmark, GEO-bench, holds about 10,000 queries drawn from nine sources, mostly informational. The main experiments ran on a simulated engine that took the top five Google results for each query and wrote answers with gpt-3.5-turbo, with a smaller test on Perplexity.ai. The authors report that the effective methods varied by domain. They also state their limits: the methods "may need to adapt over time" as engines change, they did not evaluate how the methods affect search rankings, the benchmark needs continuous updates as queries shift, and some labels may reflect subjective judgement. Read the paper as useful evidence about one controlled setup. We do not quote its headline figures here, because a number stripped of those conditions misleads more than it informs. ## What to do if you have limited time This checklist is our recommendation, with no guarantee attached. 1. Check your key pages are indexed and can show a snippet, which is Google's stated condition. Look for stray `noindex` or `nosnippet` tags. 2. Read your robots.txt and confirm it is not blocking crawlers you want, such as OAI-SearchBot, PerplexityBot or Claude-SearchBot. 3. Rewrite the headings on your most important pages as the questions buyers ask, with a direct answer in the first sentence or two beneath. 4. Keep facts consistent across your site and your external profiles: names, descriptions, locations, what you offer. 5. If you use structured data, make sure it matches what the page visibly says. 6. Start tracking whether you are named or cited for your buyers' questions. Most small teams can work through the first five in a sitting. The sixth is ongoing. ## How to tell if it is working This is harder than rank tracking. Answers vary by phrasing, by product and over time, so a single check proves little. For Google, its guide points to the Generative AI performance report in Search Console as the way to see how your content performs in its generative AI features. For other assistants, list the questions your buyers really ask, run them on a regular schedule, and record whether you are named or cited and whether the details are right. Over time that gives you a trend rather than a snapshot. Our [measurement section](/guide/#measurement) sets out a simple routine. If nothing moves, the work is still useful: the same groundwork supports ordinary search. ## Where GEO fits in GEO, generative engine optimisation, is the name the Aggarwal paper gave its framework, and Google's guide mentions it alongside AEO. In our observation the two are often used interchangeably, though some writers draw a line between them. For the longer treatment, see [what GEO is](/guide/#what-is-geo). For practical purposes the labels matter less than the work underneath. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once the groundwork is done, it shows whether ChatGPT, Claude, Gemini and Perplexity actually name and cite you for your buyers' questions. ## Questions people ask ### What is the difference between SEO and AEO? SEO is the wider work of making a site findable, useful and credible in search. AEO is a looser label for making pages easy for answer engines to quote or name. They share most of their foundations. ### Is AEO replacing SEO? Nothing in the platform documentation we read says so. Google states that SEO best practices remain relevant for its AI features and advises prioritising effective SEO over what it calls AEO/GEO hacks. ### Do I need schema markup or special files for AEO? For Google, no. Its guidance says structured data is not required for generative AI search, there is no special schema.org markup to add, and you do not need AI text files or new machine readable files to appear. ### Is AEO the same as GEO? There is no formal definition of either. In our observation many writers use them interchangeably while others draw a line, so ask anyone using the terms what they mean. ### How do I measure whether AEO is working? For Google, use the Generative AI performance report in Search Console, which Google's guide points to. For other assistants, run your buyers' real questions on a schedule, record whether you are named or cited and whether the details are right, and look for a trend over time. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central: Optimizing your website for generative AI features on Google Search](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [OpenAI: Overview of OpenAI crawlers](https://developers.openai.com/api/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: AI crawlers and robots.txt: which bots exist, and what to allow or block Canonical: https://whatisgeo.app/ai-crawlers-robots-txt/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > robots.txt lets you set a separate rule for each AI crawler. Vendors run different bots for training, search and user-requested fetches, so blocking one rarely blocks the others, and obeying the file is voluntary. There is no single switch for "AI". OpenAI, Anthropic, Perplexity and Google each document separate crawlers for separate jobs, and each name gets its own group in your robots.txt. So decide bot by bot, based on what each one is for, then check your logs to see what actually arrives. ## One file, a separate decision per bot Asking whether you want AI bots is too broad to answer. Ask what a particular bot does with your pages and whether you want that. A training crawler, a search crawler and a fetcher acting for one user asking one question are three different trades. Two caveats before the detail. First, the bot names and behaviour below are what each vendor's own page said when we checked on 18 October 2026. Those pages change, and they are the authority, so treat this page as a snapshot and check them. Second, allowing a bot only makes you crawlable. OpenAI and Perplexity recommend allowing their search bots if you want to appear in their results, but none of the pages we read promises a citation in return. ## How robots.txt works, and what it cannot do robots.txt is the Robots Exclusion Protocol, standardised as RFC 9309. The file must be called `/robots.txt` (all lowercase) and sit at the top level of a host, and it applies to that host only. Anthropic's help article makes the practical point: add the rules to every subdomain you want covered. The file is made of groups. Each group starts with one or more `User-agent` lines naming a crawler, followed by `Allow` and `Disallow` rules for paths. Under RFC 9309, crawlers match their name case-insensitively, and when several rules match a URL, the most specific one (the longest path) wins. What it cannot do matters as much: - **It is not access control.** RFC 9309 says the rules "are not a form of access authorization". Google's guide says robots.txt "cannot enforce crawler behavior" and that it is up to each crawler to obey. - **It is public.** The RFC's security section warns that listing paths in robots.txt exposes them and makes them discoverable. Do not use it to hide a private area. Put a login on it. - **It controls crawling, not indexing.** Google says a page blocked in robots.txt can still be indexed if other pages link to it. To keep a page out of Google, use `noindex` or password protection. ## The three jobs AI bots do: training, search and user-requested fetching 1. **Training.** The crawler collects content that may be used to train the vendor's models. 2. **Search.** The crawler builds an index the vendor's assistant uses to find and link to pages in its answers. 3. **User-requested fetching.** A person asks a question or pastes a link, and the assistant fetches that page on the spot. These fetchers do not crawl the web automatically. Here is how the documented names map to those jobs: | Vendor | Training | Search | User-requested fetch | | --- | --- | --- | --- | | OpenAI | GPTBot | OAI-SearchBot | ChatGPT-User | | Anthropic | ClaudeBot | Claude-SearchBot | Claude-User | | Perplexity | None documented; Perplexity says PerplexityBot is not used to crawl for foundation models | PerplexityBot | Perplexity-User | | Google | Google-Extended (a control token, also covers Gemini grounding) | Googlebot, which covers Search including AI Overviews and AI Mode | Not covered here | Blocking a training bot says nothing about search. Blocking a search bot takes you out of that vendor's index even if you were happy to be there. ## Which crawlers each vendor documents ### OpenAI OpenAI says each of its robots.txt settings "is independent of the others". - **OAI-SearchBot** surfaces websites in ChatGPT's search features. Sites that opt out "will not be shown in ChatGPT search answers, though can still appear as navigational links". OpenAI says changes can take around 24 hours to take effect. - **GPTBot** crawls content that may be used to train OpenAI's foundation models. Disallowing it signals that your content should not be used for training. - **ChatGPT-User** visits pages when a user's request needs them. OpenAI says it does not crawl automatically, that "robots.txt rules may not apply", and that it is not used to decide what appears in search. - **OAI-AdsBot** only visits landing pages submitted as ads on ChatGPT. OpenAI publishes IP lists for OAI-SearchBot, GPTBot and ChatGPT-User. ### Anthropic - **ClaudeBot** collects content that could be used for training. Blocking it signals that your future content should be left out of training data. - **Claude-SearchBot** indexes content to improve Claude's search results. Anthropic says blocking it may reduce your visibility in those results. - **Claude-User** fetches pages when a Claude user asks a question. Anthropic says blocking it stops Claude retrieving your pages for those requests, which may reduce your visibility for user-directed search. Anthropic also says its bots honour the non-standard `Crawl-delay` rule, and it publishes a list of its crawler IP addresses. ### Perplexity - **PerplexityBot** surfaces and links websites in Perplexity's search results. Perplexity says it is not used to crawl content for AI foundation models. - **Perplexity-User** fetches pages when a user asks a question. Perplexity says that because a user requested the fetch, it "generally ignores robots.txt rules". Perplexity publishes IP lists for both. ### Google **Google-Extended** is a robots.txt token, not a separate crawler: Google crawls with its usual user agents and reads the token as a control. It decides whether content Google crawls may be used to train future Gemini models and for grounding in Gemini Apps and Grounding with Google Search on Vertex AI. Google says it does not affect inclusion in Google Search and is not a ranking signal. AI Overviews and AI Mode are part of Search. Google's AI features guide says robots.txt rules for Googlebot are the control for Search crawling. To limit what Search shows from your pages, including in AI features, use `nosnippet`, `data-nosnippet`, `max-snippet` or `noindex`. ### Microsoft Bing Bing's robots.txt guide covers Bingbot and lists no separate AI crawler name. In a September 2023 announcement about Bing Chat, Microsoft used robots meta tags rather than a robots.txt name: pages tagged `NOARCHIVE` would not be included in Bing Chat answers or used to train its models, and tagged pages would still appear in Bing search results. That post is from 2023, so check Bing's current documentation before relying on it. ## Sample robots.txt patterns (and what each one does) These show the logic. Check every name against the vendor pages before you copy anything. **Opt out of training but stay available to search bots.** ``` User-agent: GPTBot User-agent: ClaudeBot User-agent: Google-Extended Disallow: / User-agent: OAI-SearchBot User-agent: Claude-SearchBot User-agent: PerplexityBot Allow: / ``` RFC 9309 allows several `User-agent` lines in one group. The first group asks the three training controls to stay off the whole site. The second states plainly that the search bots are welcome. This shows how the controls work independently. It is not advice to block training; that choice is yours. **Slow one crawler down.** ``` User-agent: ClaudeBot Crawl-delay: 1 ``` `Crawl-delay` is not part of RFC 9309. Anthropic says it supports it, but do not assume other crawlers do. ## Common mistakes: blocking a search bot when you meant to block training, and wildcard rules **Blocking by vendor instead of by job.** If you block every Anthropic or OpenAI name because you object to training, you also block the search bot and may drop out of that assistant's search results. Block the training name, and leave the search name alone if you want to be citable. **Assuming the wildcard still applies.** Under RFC 9309, a crawler obeys the `User-agent: *` group only if no group names it. Once a bot has its own group, it ignores the `*` group completely. Bing's guide says the same about Bingbot. So in this file: ``` User-agent: * Disallow: /checkout/ User-agent: OAI-SearchBot Allow: / ``` OAI-SearchBot follows only its own group and may crawl `/checkout/`. If you want it kept out, repeat the `Disallow` line in its group. **A blanket block under the wildcard.** `User-agent: *` with `Disallow: /` also blocks Googlebot and Bingbot, which takes you out of ordinary search as well as AI Overviews and AI Mode. **Misspelt names.** A name with a typo matches no crawler, so the bot you meant falls back to the `*` group, or to no rules at all if there is none. Nothing warns you. **Expecting robots.txt to stop user-requested fetchers.** OpenAI says its rules "may not apply" to ChatGPT-User, and Perplexity says Perplexity-User generally ignores them. ## Checking that your rules work: logs, testing and verifying bot identity These are our recommendations; the vendors document the details they rely on. 1. **Load the file.** Open yourdomain.com/robots.txt in a browser and on each subdomain you care about. 2. **Validate it.** Bing Webmaster Tools has a robots.txt tester that shows how a given crawler would read your rules. 3. **Read your logs.** Server or CDN logs record each request's user agent. Search for the names above, see which URLs they request, and check whether a bot you blocked keeps fetching pages. OpenAI says its bots may add a `robots.txt` marker to the user agent when fetching the file itself. Google-Extended never appears in logs because it has no user agent of its own. 4. **Verify identity.** A user agent is just text and can be faked. OpenAI, Anthropic and Perplexity publish IP lists to check requests against. Google says its crawlers can be identified by user agent, source IP and reverse DNS. 5. **Give it time.** OpenAI mentions around 24 hours for search. Google says recrawling can take days to months. For turning this into a regular routine, see [measuring progress](/guide/#measurement). ## Blocking at server or CDN level when robots.txt is not enough If logs show a bot ignoring your file, or you have content that must not be fetched at all, robots.txt is the wrong tool. Firewall, server or CDN rules actually refuse requests. Perplexity's documentation shows how to write firewall rules that match both the user agent and the published IP ranges. It uses them to allow its bots, but the same matching works for blocking. Two cautions. Anthropic says blocking its IP addresses may not reliably opt you out, because it stops its bots reading your robots.txt, so keep the robots.txt rule as well. And bot protection can cut both ways: Google's AI features guide lists making sure your CDN or hosting allows crawling among its basics, and OpenAI and Perplexity ask you to allow their search bots' IP ranges. Check that a security setting is not turning away the search bots you meant to welcome. The home guide's [crawler access](/guide/#crawler-access) section has the short version. ## How to decide: a short framework for small businesses This is our recommendation, not a vendor requirement. 1. **Write down what you want.** Do you want to be findable and citable in AI assistants? Do you object to your content being used for training? 2. **Map each bot to its job** using the table above and the vendor pages. 3. **Allow the search bots if you want to be citable.** That makes you eligible to be found. It guarantees nothing. For the platform detail, see [how to rank in ChatGPT](/how-to-rank-in-chatgpt/) and [how to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/). 4. **Block training names only if that is your decision**, one name per line. 5. **Remember the fetchers.** Some user-requested fetchers do not follow robots.txt, so anything truly private needs a login. 6. **Check your logs**, then escalate to firewall or CDN rules only if a bot ignores you. 7. **Revisit now and then.** Vendors add and rename bots, and their pages say so first. The reasoning behind this approach is in our [GEO principles](/guide/#principles). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once the search bots can reach you, it tracks whether ChatGPT, Claude, Gemini and Perplexity actually name and cite your company for your buyers' questions. ## Questions people ask ### Does robots.txt stop AI crawlers? Only crawlers that choose to obey it. RFC 9309 says robots.txt rules are not a form of access authorisation, and some user-requested fetchers, such as Perplexity-User, say they generally ignore it. To actually refuse requests you need server, firewall or CDN rules. ### What is the difference between GPTBot and OAI-SearchBot? OpenAI says GPTBot crawls content that may be used to train its foundation models, while OAI-SearchBot surfaces websites in ChatGPT's search features. Each is controlled independently in robots.txt, so you can block one and allow the other. ### If I block GPTBot, will ChatGPT stop mentioning my site? Not according to OpenAI's documentation. Blocking GPTBot signals your content should not be used for training; ChatGPT search relies on OAI-SearchBot, which you control separately. Nobody can guarantee ChatGPT will mention you either way. ### Does blocking Google-Extended remove me from Google AI Overviews? No. Google says Google-Extended does not affect inclusion in Google Search, and AI Overviews and AI Mode are part of Search. Google points to Googlebot rules and the nosnippet, data-nosnippet, max-snippet and noindex controls for those features. ### How can I check whether AI bots are crawling my site? Search your server or CDN logs for the user-agent names the vendors document, and note which URLs they request. User-agent text can be faked, so check the request's IP address against the lists OpenAI, Anthropic and Perplexity publish. ## Sources - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Google Search Central: Introduction to robots.txt](https://developers.google.com/search/docs/crawling-indexing/robots/intro) - [OpenAI: Overview of OpenAI Crawlers](https://platform.openai.com/docs/bots) - [Anthropic (Claude Help Center): Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Perplexity: Perplexity Crawlers](https://docs.perplexity.ai/guides/bots) - [Google: List of Google's common crawlers (Google-Extended)](https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers) - [Google: Overview of Google crawlers and fetchers](https://developers.google.com/search/docs/crawling-indexing/overview-google-crawlers) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Bing Webmaster Tools: How to create a robots.txt file](https://www.bing.com/webmasters/help/how-to-create-a-robots-txt-file-cb7c31ec) - [Bing Webmaster Blog: New options for webmasters to control usage of their content in Bing Chat (September 2023)](https://blogs.bing.com/webmaster/september-2023/Announcing-new-options-for-webmasters-to-control-usage-of-their-content-in-Bing-Chat) --- # Guide: AI Mode vs AI Overviews: what is the difference, and does it change what you do? Canonical: https://whatisgeo.app/ai-mode-vs-ai-overviews/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > AI Overviews are a summary Google adds to some ordinary results pages. AI Mode is a separate conversational search. Google's guidance for site owners is the same for both, though the answers and links can differ. The real difference is how people arrive. An AI Overview appears by itself on a results page when Google's systems decide it helps, while AI Mode is somewhere a person chooses to go, ask a question and keep following up. Everything below comes from Google's own documentation unless it says otherwise. ## AI Mode vs AI Overviews at a glance | | AI Overviews | AI Mode | | --- | --- | --- | | Where it appears | Inside an ordinary Google results page | A separate search experience | | How people get there | It appears by itself on some queries | The searcher opens it: google.com/ai, the AI Mode option after a search, or the Google app | | Follow-up questions | On mobile, Show more leads into AI Mode | Built in: ask, then keep asking | | Query fan-out | May be used | May be used; described in AI Mode's help page | | Model, as Google describes it | A customised Gemini model working with Search systems | A custom version of Gemini, with more advanced models | | To be a supporting link | Indexed and eligible to show with a snippet | The same | | nosnippet and max-snippet | Apply | Apply | | Search Console | Counted in the Web search type | Counted in the Web search type | Google's Search Central guide calls these two features "AI features" in Search, and it gives site owners a single set of advice for both. ## Where each one appears and how people reach it An AI Overview is something that happens to a search. Google's Help page describes it as "an AI-generated snapshot with key information and links to dig deeper", shown "when our systems determine that generative AI can be especially helpful". Search Central adds that AI Overviews are only shown when they are additive to classic Search, "and as such, often don't trigger." The searcher does not ask for one. AI Mode is somewhere a search goes. Google's AI Mode Help page lists three ways in: go to google.com/ai, type a question on google.com and tap AI Mode, or tap AI Mode on the Google app's home screen. From there people ask follow-up questions in an "Ask anything" bar. The two are connected on mobile. The AI Overviews Help page says you can select Show more at the bottom of an AI Overview, then type a follow-up to continue in AI Mode, and the context of your original search carries over. For each feature in more depth, see [what are Google AI Overviews](/what-are-google-ai-overviews/) and [Google AI Mode and SEO](/google-ai-mode-seo/). ## What the two share: query fan-out, links and eligibility Search Central says both features "may use a 'query fan-out' technique", which means issuing multiple related searches across subtopics and data sources to build a response. AI Mode's Help page describes it in plain terms: it divides your question into subtopics and searches for each one simultaneously. In practice your page could be found for a sub-question the searcher never typed. The [how it works](/guide/#how-it-works) section of our main guide covers this retrieval step in more detail. Eligibility is the same for both. To be eligible as a supporting link in either feature, Google says a page "must be indexed and eligible to be shown in Google Search with a snippet", and "there are no additional technical requirements." Eligibility is the minimum. It does not mean a page will be shown. Snippet controls apply to both as well. Google's robots meta tag documentation says `nosnippet` also prevents content "from being used as a direct input for AI Overviews and AI Mode", and `max-snippet` limits how much content may be used as that input. Both rules also change how the page appears in ordinary results, so check before you add one for an unrelated reason. ## What Google says may differ Search Central puts it in one sentence: "AI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary." Google's longer document on the two features goes further. Because AI Mode uses more advanced models and novel techniques, "a different set of responses and links will likely appear for the same query." Google also describes different jobs. AI Overviews help people "get to the gist" of a topic quickly. AI Mode is meant for queries that need "further exploration, reasoning, or complex comparisons." Google does not publish a mapping that predicts which links each feature will choose, so treat anyone claiming precise knowledge of the gap as working from their own observations. ## Are AI Overviews and AI Mode just Gemini? They run on Gemini models, but neither one is the Gemini app. Google's document on the two features says AI Overviews "use a customized Gemini model, which works in tandem with our existing Search systems", such as its quality and ranking systems and the Knowledge Graph. It says AI Mode uses "a custom version of Gemini." The AI Mode Help page also lists a named Gemini model that some signed-in users, including Google AI Pro and Ultra subscribers, can pick from a menu, alongside the default "Auto" or "Fast" options. So Gemini is the model family underneath, and AI Overviews and AI Mode are Google Search features built on it and tied to Google's index. One practical consequence: Google's crawler documentation says the Google-Extended robots.txt token controls use of your content for Gemini Apps and Vertex AI, and "does not impact a site's inclusion in Google Search." Blocking it will not remove you from either feature. For Search, the controls are Googlebot access and the snippet rules above. ## Do they cite the same pages? What research says, and its limits Google's own answer is that they will often differ, as quoted above. This page does not repeat third-party figures on how much the cited sources overlap, because we have not verified one study's sample and method well enough to report it. If you read such a study, check four things before acting on it: how many queries it used and what kind, which country and language, when the data was collected, and how a match was defined (same page, or just same domain). An overlap figure describes that query set at that time. It cannot tell you what will happen for your pages, and it does not override what Google says about the two features using different techniques. Your own tracking is the better guide. ## Do you need a different strategy for each? No. Google says "specific optimization isn't required for AI Overviews and AI Mode" and that the same foundational SEO practices apply. Its list includes allowing crawling in robots.txt and at your CDN or host, internal links, a good page experience, important content in text form, structured data that matches the visible text, and an up-to-date Merchant Center and Business Profile where relevant. One idea is worth designing around, and this is our recommendation, not a Google rule. Since both features may fan a query out into subtopics, cover the follow-up questions a topic naturally raises on a page, instead of only the headline query. [How to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/) sets out the steps, and they apply to AI Mode too. Nothing you do can guarantee a place in either feature. ## How to see each one in your reporting Search Central says traffic from both features is included in your overall Search Console data, in the Performance report under the "Web" search type. Google's documentation describes them as part of that data and does not describe a separate report for either. Search Console's help explains how each is counted: - **AI Overviews:** a click on an external link counts as a click. The whole overview takes a single position, and every link in it gets that same position. - **AI Mode:** a click on an external link counts as a click, and position follows the same method as a normal results page. A follow-up question counts as a new query, so its impressions and clicks are recorded against that new query. - **Search Labs:** experiments are not included for either feature. To see when and where you actually appear, [how to track AI Overviews](/how-to-track-ai-overviews/) explains the methods, and [how to measure AI visibility](/how-to-measure-ai-visibility/) covers the wider picture across AI assistants. ## Questions people ask ### What is the difference between AI Mode and AI Overviews? An AI Overview is a summary Google shows on some results pages when its systems judge it useful. AI Mode is a separate experience that people open themselves, where they ask a question and keep asking follow-ups. Google's guidance for site owners is the same for both. ### Is Google AI Overview just Gemini? No. Google says AI Overviews use a customised Gemini model that works alongside its existing Search systems, such as ranking and the Knowledge Graph. AI Mode uses a custom version of Gemini too, but both are features of Google Search and are separate from the Gemini app. ### Do AI Mode and AI Overviews show the same sources? Not necessarily. Google says the two may use different models and techniques, so the responses and links they show will vary, and that a different set of links will likely appear for the same query. ### Do I need to optimise differently for AI Mode and AI Overviews? No. Google says specific optimisation isn't required for either, and the same foundational SEO practices apply. A page needs to be indexed and eligible to show with a snippet, and even then Google does not guarantee it will appear. ### Can I turn off AI Mode or AI Overviews? Searchers cannot switch AI Overviews off, but Google's Web filter shows only text links. Site owners can use nosnippet, max-snippet, data-nosnippet or noindex to limit what both features use, though these also affect ordinary results. Blocking Google-Extended does not affect Search. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Help: AI Overviews in Google Search](https://support.google.com/websearch/answer/14901683) - [Google Search Help: AI Mode in Google Search](https://support.google.com/websearch/answer/16011537) - [Google: AI Overviews and AI Mode in Search (PDF)](https://search.google/pdf/google-about-AI-overviews-AI-Mode.pdf) - [Google Search Central: Robots meta tag, data-nosnippet and X-Robots-Tag specifications](https://developers.google.com/search/docs/crawling-indexing/robots-meta-tag) - [Search Console Help: How clicks, impressions and position are counted](https://support.google.com/webmasters/answer/7042828) - [Google crawling infrastructure: Google's common crawlers (Google-Extended)](https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers) --- # Guide: AI SEO course: what a good one teaches, and how to choose one Canonical: https://whatisgeo.app/ai-seo-course/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-23 > A good AI SEO course teaches two skills: using AI for SEO work with human review, and getting content cited in AI answers. Check its claims against platform documentation; no course can guarantee inclusion. Before you compare courses, decide which of the two skills you are paying to learn. Then judge the syllabus on whether its claims can be checked. A course that cannot show where its advice comes from is asking you to trust it, and this subject changes too often for that to be safe. ## What people mean by an AI SEO course The phrase is used for two different things, and many course pages mix them. The first is using AI tools to do SEO tasks faster: grouping keywords, drafting briefs, summarising audit data, writing first drafts. The skill here is a workflow with review built in, because the output still needs someone who knows what good looks like. Whether machine-written pages help or hurt is covered in [is AI content good for SEO](/is-ai-content-good-for-seo/). The second is optimising for AI answers: getting your pages cited, linked or described accurately in answers from Google AI Overviews and AI Mode, ChatGPT, Perplexity, Claude and others. That is what this site calls generative engine optimisation, and it is a separate job from the first. Plenty of people need both. Check which one a course actually teaches before you sign up, because a title that says "AI SEO" can mean either. ## What a good course should cover These are the topics we would expect. Most can be checked against a document the platform publishes itself. **The foundations are ordinary SEO.** Google's guidance on AI features says you can apply the same foundational SEO best practices to AI Overviews and AI Mode as to Search overall. To be eligible to appear as a supporting link, a page must be indexed and eligible to be shown in Google Search with a snippet, and Google says there are no additional technical requirements. A course that skips this, or hides it under a new acronym, is starting in the wrong place. **AI-assisted content, with its limits.** Google's 2023 guidance says appropriate use of AI or automation is not against its guidelines, while using it to generate content mainly to manipulate rankings breaks its spam policies. Its current page on generative AI content warns that generating many pages without adding value for users may breach the scaled content abuse policy, and says it is critical to fact-check and review AI-generated content before publishing. A good course teaches that review step. One that shows you how to publish hundreds of pages with no checking is teaching you to break a published policy. **Crawler access, matched to each company's own documentation.** OpenAI, Anthropic and Perplexity each publish pages describing their crawlers, and they separate crawling for search from crawling for model training. OpenAI, for example, lists OAI-SearchBot for search in ChatGPT and GPTBot for training, and says each setting is independent. Anthropic lists Claude-SearchBot, Claude-User and ClaudeBot. Perplexity says PerplexityBot surfaces sites in its search results and is not used to crawl content for AI foundation models. A crawler module that treats "the AI bot" as one thing has not read these pages. Our [guide to AI crawlers and robots.txt](/ai-crawlers-robots-txt/) links all three. **Measurement.** A course should show you how to tell whether anything changed, from referral traffic in your analytics to checking AI answers for the questions your buyers ask. See [how to measure AI visibility](/how-to-measure-ai-visibility/) for what can and cannot be measured. ## How to judge a course syllabus We are not naming a "best" course, because we have no tested method that would back a ranking. What follows is our recommendation for judging any course yourself. 1. **Ask for the syllabus before you pay.** If the provider will not share a module list, that tells you something. 2. **Check when it was last updated.** Platform documentation changes, so an undated course is a weak sign. 3. **Check that each claim names a source.** Platform rules should point to the platform's own page. "Google says" with no link is not enough. 4. **Check how research is used.** The paper that introduced the term is Aggarwal et al., "GEO: Generative Engine Optimization" (KDD 2024). The authors tested ways of rewriting content on GEO-bench, a benchmark of 10,000 queries, using a generative engine they built from the top five Google results and a language model, and also ran tests on Perplexity. They report visibility gains of up to 40% in their setup, and say the effect varied across domains. A good course gives you those conditions. One that quotes "up to 40%" as a promise for your site has misread the paper. 5. **Prefer courses that teach measurement,** including how to set a baseline before you change anything. 6. **Treat any guarantee of inclusion as a red flag** (more on this below). You cannot verify every line of a syllabus. You can check whether its claims are checkable, which is most of the work. ## Red flags that a course is selling hype **A guarantee.** No course can promise your site will appear in ChatGPT answers or Google AI Overviews. Google's own wording is about eligibility: meeting the requirements makes a page eligible to be shown, which is a long way from being shown. Nobody outside the platform controls the outcome. **A hidden method.** If the pitch is a technique the platforms supposedly do not want you to know, ask why it appears in no documentation. The platforms publish a lot about how they crawl and what they require, and a course should be able to point to it. **Confident numbers with no trail.** Statistics with no named study, no conditions and no date. Real research comes with its limits attached, as the GEO paper does. **No measurement module.** If a course cannot tell you how you would know it worked, you have no way to judge what it taught. **An "official" certificate.** We found no GEO or AI SEO certification issued by Google, OpenAI, Anthropic or Perplexity in the platform documentation we read. Course providers and vendor academies do issue their own certificates; check who issues one before treating it as a credential. ## Free courses and vendor academies: what they are good for Several software companies run free training on this topic. It is vendor material, made with the vendor's own product in view, and it can still be useful. We describe each only as its own page did when we checked it on 23 October 2026. - **HubSpot Academy** has a learning path called "SEO in the Age of AI". Its page describes six playlists on answer engine optimisation, covering how search behaviour has changed, Google, ChatGPT, Perplexity and Gemini, AI content tools, video, and implementation and measurement. The page says the path ties this to HubSpot's own Loop Marketing framework, and the Academy describes its courses as free. - **Semrush Academy** lists its courses under "Free Online Marketing Courses", including "AI Search Operating System" and "On-Page SEO and AI Search Essentials with Semrush". The second title names Semrush itself, a reminder that vendor courses are built around the vendor's tools. - **Coursera** is a marketplace with courses from many providers. A search for "ai seo" there returns courses and specialisations that were marked "Free trial" when we checked. Use free courses for vocabulary, a first look at the tools and a sense of what each company thinks matters. Do not expect a neutral answer on whether you need that company's software. Test what they teach against the platform documentation, as you would with a paid course. ## Course, tool, agency or doing it yourself A course is one option among several, and the right one depends on what you are short of. | If you are short of | Consider | Watch for | |---|---|---| | Knowledge | A course, or the platform documentation itself | Claims with no source | | Time on repetitive tasks | A tool; see [AI SEO tools](/ai-seo-tools/) | Output nobody reviews | | Capacity or in-house skill | An agency; see [choosing a GEO agency](/generative-engine-optimization-agency/) | Guarantees and vague reporting | | None of these | Doing the work from the documentation | Skipping measurement | The foundations are public. Google, OpenAI, Anthropic and Perplexity all document what they require, so a course mainly saves you time and gives you structure. ## What to do after the course Apply what you learned to one site soon, while it is fresh. Set a baseline first: record how your pages currently appear for the questions you care about, so you have something to compare against later. Then make one change at a time where you can. If you change six things at once, you will not know which one mattered. Go back to the primary sources every so often. Platform guidance is updated, and a course recorded last year will not know what has changed since. When you need to reset your footing, the [principles section of our main guide](/guide/#principles) sets out the approach this site recommends. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want to see, after a course, whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for your buyers' questions, it is one way to track that over time. ## Questions people ask ### What should an AI SEO course teach? Two skills: using AI tools for SEO tasks with human review, and getting content cited in AI answers. It should also cover crawler access as each platform documents it, and how to measure results. ### Are free AI SEO courses worth doing? They are a reasonable starting point for vocabulary and getting to know the tools. They are vendor material with the vendor's own product in view, so check what they teach against platform documentation. ### Is there an official certification in AI SEO or GEO? We found none issued by Google, OpenAI, Anthropic or Perplexity in the platform documentation we read. Course providers and vendor academies do issue their own certificates, so check who issues one before treating it as a credential. ### Can a course guarantee my site appears in ChatGPT or AI Overviews? No. Google describes its requirements as making a page eligible to appear, and nobody outside the platform controls whether it does. Treat a guarantee as a red flag. ### Do I need a course, or can I learn this from documentation? You can learn a lot from documentation alone, because the platforms publish their requirements. A course saves time and adds structure, and you should still read the primary sources. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central: Google Search's guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full text (v3)](https://arxiv.org/html/2311.09735v3) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [HubSpot Academy: SEO in the Age of AI learning path (checked 23 October 2026)](https://academy.hubspot.com/seo-in-the-age-of-ai-learning-path) - [HubSpot Academy home page (checked 23 October 2026)](https://academy.hubspot.com) - [Semrush Academy: Free Online Marketing Courses (checked 23 October 2026)](https://www.semrush.com/academy/courses/) - [Semrush Academy home page (checked 23 October 2026)](https://www.semrush.com/academy/) - [Coursera: search results for "ai seo" (checked 23 October 2026)](https://www.coursera.org/courses?query=ai%20seo) --- # Guide: AI SEO tools: what they do, which kinds exist, and how to choose Canonical: https://whatisgeo.app/ai-seo-tools/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > 'AI SEO tools' means two things: SEO software with AI features added, and tools that track how AI answers mention you. Choose by the job you need done, and check every AI output yourself. Before you compare a single product, decide which of the two groups you're shopping in and which job you need done. Most disappointment with these tools starts with buying the wrong kind for the problem you actually have. ## What people mean by AI SEO tools The phrase covers two different groups of product sold under one label. The first group is familiar SEO software: keyword research, page analysis, site crawling, content briefs. All of that existed before, and much of it now has an AI assistant added. The second group is newer. These tools put questions to AI systems such as ChatGPT, Gemini or Perplexity and record whether your business turns up in the answers, how it is described and which sources are cited. This page doesn't rank products. We have no method we could show you that would back a "best" list, so we sort tools by job instead. If you want the wider definition first, [what is AI SEO](/what-is-ai-seo/) covers it. ## The main kinds of AI SEO tool, by job Four jobs cover almost everything on the market. | Job | What the tool does | Where AI usually appears | |---|---|---| | Research | Finds the queries people use, topics you're missing and what competitors cover | Grouping queries into themes, suggesting topics, summarising data | | Content | Outlines, drafts, edits and scores pages | Drafting and rewriting copy, writing briefs | | Technical | Crawls your site for broken links, slow pages and indexing problems | Explaining issues in plain sentences, suggesting fixes | | AI-answer tracking | Asks AI systems a fixed set of questions and logs mentions and citations | The AI system is the thing being measured | Plenty of products cover two or three of these jobs. That's fine, as long as you know which job you're hiring for before you read a feature list. A tool that's good at drafting does nothing for a crawl problem, and a crawler won't tell you what AI answers say about you. ## What AI is and is not doing inside these tools In most products the AI is doing language work: summarising a pile of data, suggesting topics, drafting copy, grouping queries and turning a report into readable sentences. That saves time on the dull parts. What it doesn't have is knowledge of your business. It doesn't know which services make you money, which customers you want or which claims on your page are accurate. Google's own guidance on generative AI content makes the underlying point: generative models "don't retrieve facts, but predict a likely sequence of words". So a suggested keyword may have little real demand behind it, and a drafted paragraph may say something untrue about your product. The working rule (our recommendation): treat AI output as a fast first pass, check it, and keep a person who knows the business in the review step. A confident-sounding tool can still be wrong. ## Free tools worth starting with You can start without paying anyone, and the first tools to set up are documented by the search engines themselves. **Google Search Console.** Google describes it as a free service. Its Performance report shows clicks, impressions, click-through rate and average position, and you can break those down by the queries that brought people to your site. That is real data on how you already appear, which no third-party estimate can match. **Bing Webmaster Tools.** Microsoft's equivalent for Bing. In February 2026 Microsoft announced an AI Performance report in public preview, which shows how often your pages are cited as sources in Microsoft Copilot, AI-generated summaries in Bing and some partner integrations, plus the queries the AI used to retrieve your content. Microsoft notes that these figures show citation activity, not ranking or importance, and that the query data is a sample. We couldn't find a Microsoft page that states the sign-up terms in plain text, so check them when you register. Neither is sold as an "AI" product, and that is useful: they show what is already happening on your own site before you spend anything. Set both up first and you'll know which problem you're actually trying to solve. Paid products often have free tiers, but we haven't checked any vendor's current terms for this page, so we don't name them. If you try one, read the vendor's own pricing page and note the date you checked. ## AI content tools: what Google says and where the risk sits This is the worry that stops many small businesses, so here is what Google actually documents. Google's 2023 guidance says "appropriate use of AI or automation is not against our guidelines". What it does treat as spam is using automation, including AI, "with the primary purpose of manipulating ranking in search results". Its spam policies define scaled content abuse as generating many pages mainly to manipulate rankings and not to help users, and they list "using generative AI tools or other similar tools to generate many pages without adding value for users" as an example. So Google cares about the purpose and value of the pages, whatever wrote them. Publishing a hundred thin, near-identical pages is the risk. Using a drafting tool to speed up a page that a knowledgeable person then checks and improves is a different activity. Google's generative AI guidance also says to fact-check and review all AI-generated content before publishing, including titles, meta descriptions, structured data and image alt text. Following this guidance keeps you within Google's policies. It doesn't promise a ranking. For the fuller quality question, see [is AI content good for SEO](/is-ai-content-good-for-seo/). ## Tools that track visibility in AI answers This is the newer half of the phrase. Trackers put a set of questions to AI systems and log whether your business is mentioned, how it is described and which sources are cited. The useful part is the record. Without one you're relying on someone's screenshot; with one you can see change over time. For how to set up measurement, including a manual method, see [how to measure AI visibility](/how-to-measure-ai-visibility/), and for a closer look at this category and the questions to ask vendors, see [generative engine optimisation tools](/generative-engine-optimization-tools/). The limit, stated plainly: nobody can guarantee a ranking or a citation in AI answers, and no tool changes that. A tracker reports on what the AI system did; it doesn't control what it does next. Google's documentation on AI features says there are no additional requirements or special optimisations needed to appear in AI Overviews or AI Mode, and that you don't need new machine-readable files, AI text files or special structured data. Be wary of any product that sells a special switch for Google's AI features. There is research on how content changes affect visibility in AI answers. Aggarwal et al. ("GEO: Generative Engine Optimization", accepted to KDD 2024) built a benchmark of queries across several domains, tested content changes against generative engines in their own setup, and reported visibility gains of up to 40%, with the most effective changes varying by domain. That is a finding from a defined experimental setup and is no forecast for your site. If you want to check your own position step by step, the [AI visibility audit](/ai-visibility-audit/) guide walks through it. ## How to choose: a checklist you can run This is our recommendation, not platform guidance. 1. **Name the job.** Research, content, technical or tracking. Write it down. 2. **Start with the free data.** Set up Search Console and Bing Webmaster Tools first. 3. **Trial on your own site.** A demo on someone else's data proves little. 4. **Verify the output.** Check a sample of suggestions against what you know to be true. 5. **Keep human review.** Someone who knows the business signs off before anything goes live. 6. **Prefer exportable data.** If you can't take your data with you, you lose your history when you leave. 7. **Check the vendor's own pages.** For prices and terms, use the vendor's site and note the date you looked. 8. **Ask what they can't promise.** A vendor guaranteeing rankings or citations is promising something nobody controls. ## Common mistakes when buying AI SEO tools **Buying before naming the job.** You end up with a drafting tool when the real issue was a crawl error. **Trusting a "best of" list with no stated method.** If you can't see how the ranking was made, you can't weigh it. **Publishing unchecked output.** Fast and wrong is still wrong, and Google tells you to review AI content before it goes live. **Expecting a guarantee.** Some marketing implies one. No tool can deliver it. **Skipping the free tools.** Search Console already shows your queries, clicks and impressions. Paying for a dashboard before looking at that is backwards. **Skipping a baseline.** Without one, you can't tell whether anything worked. The [measurement section of our guide](/guide/#measurement) is a good place to start. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you need the AI-answer tracking job done for more questions than you can check by hand, it is one option for tracking whether ChatGPT, Claude, Gemini and Perplexity name and cite your company; put the checklist questions on this page to it as to any other tool. - [ScriptGrain](https://scriptgrain.com/): If the job is drafting content, it drafts in your own measured voice, and the same rule applies: a person who knows the business checks and edits every draft before it goes live. ## Questions people ask ### What are AI SEO tools? The label covers two groups: classic SEO software that has added AI features for research, content and technical work, and newer tools that track whether and how AI answers mention your business. Pick by the job you need done, not by the label. ### Are there free AI SEO tools worth using? The most useful free starting point isn't sold as an AI product. Google says Search Console is a free service, and its Performance report shows your clicks, impressions, click-through rate and position by query. Microsoft's Bing Webmaster Tools is worth adding too, and its AI Performance report (in public preview) shows citations in Copilot and Bing's AI summaries. ### Do AI SEO tools get you ranked or cited in AI answers? No tool can guarantee that. Google says there are no additional requirements or special optimisations needed to appear in AI Overviews or AI Mode, so a tool can help you work and report, but it can't control the outcome. ### Is it safe to use AI to write SEO content? Google says appropriate use of AI or automation is not against its guidelines. Using automation mainly to manipulate rankings is spam, and generating many pages without adding value for users is listed under scaled content abuse. Google also tells you to fact-check and review AI-generated content before publishing. ### What should a small business look for in an AI SEO tool? Our recommendation: one clearly named job, a trial on your own site, output you can verify, a person reviewing before anything goes live, and data you can export. Treat any promise of guaranteed rankings or citations as a warning sign. ## Sources - [Google Search Central: Google Search's guidance on generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central: Spam policies (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Search Console Help: About Search Console](https://support.google.com/webmasters/answer/9128668) - [Search Console Help: Performance report (Search results)](https://support.google.com/webmasters/answer/7576553) - [Bing Webmaster Blog: Introducing AI Performance in Bing Webmaster Tools Public Preview](https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: AI SEO vs traditional SEO: what is the difference, and which meaning do you need? Canonical: https://whatisgeo.app/ai-seo-vs-traditional-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-29 > "AI SEO" means two things: using AI tools to do SEO work, or optimising to appear in AI answers. Traditional SEO is the foundation for both, and neither route can guarantee results. The difference depends on which meaning is in play: AI-assisted SEO changes only how the work gets done, while optimising for AI answers changes where you want your pages to appear. In both cases the platforms' own rules point back to the foundations traditional SEO already covers, though each meaning rests on different evidence and carries different risks. On this page each claim is labelled as **documented** (a platform's own documentation says so), **research** (a named study, with its limits) or **our recommendation** (sensible practice, nothing more). ## What "AI SEO" can mean: two different things **Meaning one is about process.** You use AI tools to help with SEO work: research, drafting, grouping keywords, auditing pages. The goal is the same as before. Only the tools have changed. **Meaning two is about destination.** You want your pages found, used and cited when someone gets an answer from an AI system such as Google's AI Overviews and AI Mode, ChatGPT, Claude or Perplexity. Other names for this include GEO (generative engine optimisation) and AEO (answer engine optimisation). Much of the confusion comes from mixing the two. A claim that "AI SEO saves time" is about meaning one. A claim that "AI SEO wins citations" is about meaning two. Each needs different evidence, so separate them before you weigh either. The debate over names is covered in [what AI SEO is](/what-is-ai-seo/) and [SEO vs GEO vs AEO](/seo-vs-geo-vs-aeo/). ## Traditional SEO in one paragraph: what it is trying to achieve Traditional SEO makes a site easy for search engines to crawl, index and understand, and worth showing for the searches it targets. Google sums up the foundations as meeting its technical requirements for Search, following its spam policies, and creating helpful, reliable, people-first content (documented). The aim is to be the result a searcher chooses. Everything else is method. ## Meaning one: using AI tools to do SEO work You are still doing SEO, with AI doing some of the work. Google has said clearly how it views that. - **Using AI is allowed.** Google's guidance on generative AI content says these tools can be useful for researching a topic and adding structure to original content (documented). - **Scale without value is spam.** Google's spam policies define scaled content abuse as generating many pages "for the primary purpose of manipulating search rankings and not helping users", and say this applies "no matter how it's created". Using generative AI to produce many pages without adding value for users is one of the listed examples (documented). - **Review is expected.** Google warns that generative models predict a likely sequence of words and do not retrieve facts, so their output may contain inaccuracies. It calls manual fact-checking and review of all AI-generated content before publishing "critical", and says this also covers title elements, meta descriptions, structured data and image alt text (documented). So the risk with meaning one is publishing a lot of thin or wrong material because it was cheap to make. That risk existed before AI, but the tools make it faster to produce that kind of material. There is more detail in [is AI content good for SEO](/is-ai-content-good-for-seo/). Cost, ROI and speed figures attached to AI SEO tools usually come from the companies selling them. Treat them as vendor claims. This page repeats none of them. ## Meaning two: optimising to be found and cited in AI answers ### Google AI Overviews and AI Mode Google's AI features page is the clearest primary source here (documented): - You can apply the same foundational SEO best practices to AI features as to Google Search overall. - To be shown as a supporting link, a page must be indexed and eligible to be shown in Google Search with a snippet. Google says there are "no additional technical requirements". - Specific optimisation "isn't required" for AI Overviews and AI Mode, though existing SEO fundamentals "continue to be worthwhile". - AI Overviews only appear when Google's systems judge them to add something to classic Search, and they "often don't trigger". - robots.txt rules for Googlebot control how your site is crawled for Search. Google-Extended is a separate control for training and grounding in some of Google's other systems. ### ChatGPT, Claude and Perplexity Outside Google, the main thing the platforms document is crawler access. Each one names separate crawlers for different jobs (documented): | Platform | Search crawler | Training crawler | User-triggered fetcher | |---|---|---|---| | OpenAI | OAI-SearchBot | GPTBot | ChatGPT-User | | Anthropic | Claude-SearchBot | ClaudeBot | Claude-User | | Perplexity | PerplexityBot | None listed; Perplexity says PerplexityBot is not used to crawl content for AI foundation models | Perplexity-User | OpenAI says each robots.txt setting is independent. Sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. Anthropic says its bots honour robots.txt, and that blocking Claude-SearchBot or Claude-User may reduce your visibility in Claude's search results. Perplexity says Perplexity-User generally ignores robots.txt because a user asked for the page. Setup is covered in [AI crawlers and robots.txt](/ai-crawlers-robots-txt/). ### What the research adds The main academic reference is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024 (research). The authors built GEO-bench, a benchmark of 10,000 queries. They tested ways of rewriting source content on their own generative engine, which wrote answers from the top five Google results for each query. Adding citations, quotations from relevant sources and statistics raised visibility by up to 40% on the measures the authors defined. They also ran the methods on Perplexity.ai. The limits matter. The engine, benchmark and measures were the authors' own, the paper says how well the methods worked varied by subject area, and live AI products change over time. Treat the paper as a guide to where to experiment. It does not show that any change will work on a given platform today. ## Side-by-side comparison of traditional SEO, AI-assisted SEO and optimising for AI answers | | Traditional SEO | AI-assisted SEO (meaning one) | Optimising for AI answers (meaning two) | |---|---|---|---| | What it is | A discipline with a target | A way of working | A target | | Goal | Rankings and clicks from search results | The same goal, with AI tools in the workflow | Being used, cited or described accurately in AI answers | | Platform rules that apply | Search technical requirements, spam policies, helpful content | The same, plus Google's guidance on AI content and scaled content abuse | Google: the same as Search. Others: crawler access through robots.txt | | Main risk | Pages that are not indexed or not useful | Volume replacing value, and errors nobody checked | Blocking the wrong crawler, or expecting control no platform offers | | How you see progress | Search Console, rankings, traffic | The same as traditional SEO | Google: Search Console's Generative AI performance report shows impressions in AI Overviews and AI Mode (clicks stay within the overall Web search figures). Other assistants: you check the answers yourself or with a tracking tool | The middle column changes only the method. The real difference is in the last column: a different target, and a less direct way to see whether you are reaching it. ## What stays the same across all three Pages that can be crawled and indexed. Content that is accurate and answers the question properly. Internal links a crawler can follow. Important content available as text, and structured data that matches what is visible on the page. Google lists each of these as still worthwhile for its AI features (documented). If you have been doing traditional SEO well, that work carries straight over. If you have not, adding AI on top will not make up for it. The home guide's [GEO vs SEO section](/guide/#geo-vs-seo) covers where they overlap in more depth. ## What is documented, what is research, and what is practice | Kind of claim | Examples on this page | How much weight to give it | |---|---|---| | Documented | Google's AI features page, its AI content guidance, spam policies and Search Console help; the OpenAI, Anthropic and Perplexity crawler pages | The platform's own statement. Check the page date, because these change | | Research | Aggarwal et al., KDD 2024 | Evidence under stated conditions, with stated limits | | Practice | The order of work suggested below | A recommendation. Useful, never a guarantee | If a claim fits none of these, such as an unattributed percentage on a sales page, treat it as marketing. ## Which one should you spend time on first? We recommend the same order in both situations: the foundations first, then your workflow, then visibility in AI answers. 1. **Small business with little time:** check that your key pages are indexed and can show a snippet, then make them genuinely useful. Use AI tools only where a person reviews the output before it goes live. Make one deliberate decision about which AI crawlers your robots.txt allows. 2. **In-house team with some capacity:** keep traditional SEO as the base. Use AI tools to speed up research and drafting, with editorial review as a fixed step. Then start checking how AI assistants answer your buyers' questions, using [how to measure AI visibility](/how-to-measure-ai-visibility/) as the method. ## What nobody can promise Google's AI features page explains what makes a page eligible. It does not promise inclusion, and AI Overviews often do not appear at all. None of the crawler pages from OpenAI, Anthropic or Perplexity says that allowing their bots will get you cited. The GEO research reports results on its own benchmark and tells you nothing certain about your site. So anyone promising citations, a fixed timeline or a return on investment is promising something the platforms themselves do not. What you can do is make your pages eligible, accessible and useful, test changes carefully, and read research with its limits in mind. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once the foundations are in place, it checks whether ChatGPT, Claude, Gemini and Perplexity actually name and cite your company for your buyers' questions. ## Questions people ask ### Is AI SEO replacing traditional SEO? Not according to the platforms. Google says the same foundational SEO best practices apply to AI Overviews and AI Mode, with no additional technical requirements, so traditional SEO stays the base for both meanings of AI SEO. ### Is AI SEO the same as GEO or AEO? Only in one of its two meanings. GEO and AEO are names for optimising to appear in AI answers, while "AI SEO" is also used for doing SEO work with AI tools. ### Do I need to stop doing traditional SEO to appear in AI answers? No. Google says a page must be indexed and eligible to show with a snippet to appear as a supporting link in AI Overviews or AI Mode, and those are traditional SEO basics. ### Is content written with AI bad for SEO? Google says generative AI can be useful for research and structure, but using it to produce many pages without adding value for users may count as scaled content abuse under its spam policies. Google also says all AI-generated content should be fact-checked and reviewed by a person before it is published. ### How can I tell whether AI search is sending my site any visibility? For Google, Search Console's Generative AI performance report shows how often links to your site appeared in AI Overviews and AI Mode. It reports impressions only; clicks from those features are counted within the Performance report under the Web search type. For ChatGPT, Claude, Perplexity and others, you have to check the answers to your buyers' questions yourself or with a tracking tool. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Search Console Help: Generative AI performance report (Search)](https://support.google.com/webmasters/answer/16984139) - [Google Search Central: Guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central: Spam policies for Google web search](https://developers.google.com/search/docs/essentials/spam-policies) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization (full text, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: AI tools for SEO keyword research: what they can do, what they cannot, and how to use them Canonical: https://whatisgeo.app/ai-tool-for-seo-keyword-research/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-30 > AI tools are good at generating, grouping and sense-checking keyword ideas, but a chatbot has no search volume data of its own. Use AI for ideas and clustering, and check demand in Google's own tools. Keyword research is really three jobs, and AI suits two of them. Coming up with ideas and grouping them are language tasks a model does well. Demand data has to come from a source that records real searches, such as Google Keyword Planner or your own Search Console account. ## Can AI do keyword research? The direct answer by job | Job | What it means | Can an AI model do it? | |---|---|---| | Idea generation | Listing the phrases and questions a buyer might use | Yes, well, from a good brief | | Clustering and intent | Grouping phrases into topics and judging what each searcher wants | Yes, as a first pass that you check | | Demand data | How many people search, how competitive it is, whether interest is rising | Not on its own; this needs a real data source | The split follows from how the models work. Google's guidance on generative AI content says generative models "don't retrieve facts, but predict a likely sequence of words based on their training data", and that their output may contain inaccuracies. Writing plausible phrases is exactly that kind of task. Counting searches is a different one. ## What a chatbot cannot tell you: volume, difficulty and trends Ask a chatbot how many people search for a phrase each month and it will often give you a confident number. On its own, the model has no search log to read that number from, so treat it as a guess. That is our reasoning from how the models work and from Google's description above, not a measured finding. Difficulty and trends have the same problem. Without a live data source, a model cannot see who ranks today, how strong those pages are, or whether interest is climbing or fading. It can describe what a competitive topic usually looks like. It cannot measure yours. Some chat tools can search the web, and some SEO products add AI features on top of their own databases. In both cases a figure is only as good as its origin. Ask the tool, or the vendor, which data source produced the number. A source you can check is an answer; a vague reply is a warning sign. ## Where AI genuinely helps: ideas, clustering and intent Google's guidance says generative AI "can be particularly useful when researching a topic, and to add structure to original content". Keyword research benefits from both. The uses below are our recommendations. - **Ideas.** Give the model your product, your customer and a few seed phrases, then ask for variations in the words buyers use, including the clumsy ones. It will surface angles that the tidy language on your own website misses. - **Clustering.** Paste in a long list and ask it to group the phrases by topic. A model produces a respectable first pass far faster than you could sort by hand. Read the groups yourself, though: it will sometimes merge two things that need separate pages. - **Intent.** Ask what each group is after: learning, comparing or buying. Someone who wants a definition and someone who wants a quote need different pages. Check the model's read by searching the phrase and looking at what already ranks. ## Where the real numbers come from ### Google Keyword Planner Keyword Planner is part of Google Ads. Google's help page says it lets you discover new keywords related to your business and see estimates of how many searches they get each month and what it costs to target them with ads. You can start from words or a website ("Discover new keywords"), or paste in a list you already have ("Get search volume and forecasts"). The same page makes three points worth knowing before you start: - To use even basic features such as getting keyword ideas, you must complete your Google Ads account setup, including entering billing information. - Keywords with very low search volumes, or ones considered sensitive, cannot be discovered or forecast. Google says its policies often prevent results for sensitive categories such as health or finance. - The tool is built to research keywords for Search ad campaigns, so its figures are estimates made for ad planning. Keyword Planner can also show keyword ideas grouped by theme, and can automatically organise keywords into ad groups (in English only). That gives you a second opinion on the clusters your AI model suggested. ### Search Console The Search Console Performance report shows how your own site performs in Google Search. Its Queries view lists the searches your site appeared for, with clicks, impressions, click-through rate and average position. The default view covers the past three months. Google notes that a query appearing in the list does not mean you will see your site when you search it yourself, because results depend on the time, place, device and recent history of the person searching. This is real behaviour on your own pages, which makes it the most reliable check you have. Its limit is scope: it covers only your site, and only queries where you already appeared, so it says nothing about demand in the wider market. ### Your own records Sales calls, support emails, reviews and enquiry forms hold the exact words your customers use. No tool has that data, and it costs nothing to collect. ## A simple workflow you can run yourself This method needs a chatbot, Keyword Planner and Search Console. Remember that Keyword Planner needs a Google Ads account with billing information entered. The steps are our recommended method, not a platform requirement. 1. Write down your product, your customer and the problem you solve, in plain words. 2. Ask an AI model for a long list of phrases and full questions a buyer might use. 3. Ask it to group the list into topics and label the likely intent of each group. 4. Paste the strongest phrases into Keyword Planner's "Get search volume and forecasts" and note the estimates. Expect some phrases to return nothing. 5. Open the Search Console Performance report, choose Queries, and look for phrases where you already get impressions but few clicks. 6. Pick a short list and search each phrase yourself before you write. If every top result is a format you cannot match, it is better to find that out before you start. For other no-cost options, see [free AI tools for SEO](/free-ai-tools-for-seo/). ## Keyword research for AI answers: questions and prompts, not just keywords People type short phrases into a search box, but they often ask an AI assistant a full question with context attached. "Which accountant suits a small bakery that has just hired its first employee?" carries far more detail than "bakery accountant". We recommend adding a second layer to your research to cover questions like that. Google's documentation on AI features adds a relevant detail. AI Overviews and AI Mode may use a "query fan-out" technique, issuing multiple related searches across subtopics and data sources to build a response. So a single question can touch several subtopics, which is a good reason to cover the follow-up questions around a topic as well as the main phrase. The same page says that to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to appear in Google Search with a snippet, and that there are no additional technical requirements. To build a prompt list (our recommendation): - Collect the questions customers already ask you, in their own words. - Ask an AI model what else a buyer in that situation might ask. Cut anything you have never heard a real customer say. - Put a handful of those questions to the assistants your buyers use, and note whether you are named and which sources are cited. Repeat this over time, because answers vary. - Write pages that answer the questions clearly; see [how to write content for AI search](/how-to-write-content-for-ai-search/). Research on this exists but is young. Aggarwal et al. (GEO, KDD 2024) built a benchmark of queries across multiple domains and tested ways of making content more visible in generative engine responses. They found that how well each method worked varied by domain. They also describe generative engines as black-box and fast-moving, and say content creators have little to no control over when and how their content is shown. Nobody can guarantee inclusion in an AI answer. To track how you appear, see [how to measure AI visibility](/how-to-measure-ai-visibility/) and the [measurement section of our main guide](/guide/#measurement). ## Kinds of tool compared This table compares kinds of tool, not products, and ranks none of them. | Kind of tool | What it is for | What data it uses | What to check | |---|---|---|---| | General AI chatbot | Ideas, clustering, intent labels, question lists | Patterns from its training; web pages only if it searched | Treat any volume it quotes without a named source as a guess | | Google Keyword Planner | Keyword ideas and monthly search estimates | Google's own data, inside Google Ads | Needs account setup with billing information; low-volume and sensitive terms may not appear | | Google Search Console | Queries your site already appears for | Your own clicks, impressions, click-through rate and positions | Covers your site only, not the wider market | | SEO suite with AI features | A combined research workflow | The vendor's own data, which varies by product | Ask where the volume figures come from and how they are estimated | For more on using ChatGPT itself in SEO work, see [ChatGPT SEO tools](/chatgpt-seo-tools/). ## Mistakes to avoid - **Trusting a chatbot's volume figure.** Check it in Keyword Planner or ignore it. - **Publishing a page per keyword from raw AI text.** Google's spam policies give "using generative AI tools or other similar tools to generate many pages without adding value for users" as an example of scaled content abuse. Its generative AI guidance also says to fact-check and review AI-generated content before publishing. What matters is whether a page adds value for readers, however it was produced; [is AI content good for SEO](/is-ai-content-good-for-seo/) covers this in detail. - **Skipping the results check.** The pages already ranking show you what format and depth a phrase needs. - **Researching keywords only.** Buyers who ask an assistant use full questions, so keep a question list alongside your keyword list. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If your list of buyer questions is longer than you can check by hand, it tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for those questions. ## Questions people ask ### Can ChatGPT do keyword research? Partly. It is good at suggesting phrase ideas, grouping them and judging intent. On its own it has no search volume data, so check demand in Google Keyword Planner and your Search Console account. ### Does an AI keyword tool give accurate search volume? Only if it draws on a real data source, and even then the figures are estimates. If a tool quotes volume, ask where the number comes from and compare it with Keyword Planner. ### Is there a free AI tool for keyword research? Any general chatbot can do the idea and clustering work, including free versions where they exist. For demand data, use Google Keyword Planner, which needs a Google Ads account with billing information entered, and your own Search Console data. ### How do I find the questions people ask AI about my business? Start with your own records: sales calls, support emails, reviews and enquiries. Keep questions an AI model suggests only if they match what real customers say. Then put a sample to the assistants your buyers use and note whether you are named and cited. ### Should I use AI to cluster keywords? Yes, as a first pass. Read every group before you act on it, because a model can merge topics that need separate pages. If you use Keyword Planner, compare its grouping with the model's. ## Sources - [Google Ads Help: Use Keyword Planner](https://support.google.com/google-ads/answer/7337243?hl=en) - [Search Console Help: Performance report (Search results): overview and basic setup](https://support.google.com/webmasters/answer/7576553?hl=en) - [Google Search Central: Google Search's guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central: Spam policies for Google web search (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: AI visibility audit: what to check, step by step, and how to run one yourself Canonical: https://whatisgeo.app/ai-visibility-audit/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > A free AI visibility audit: check crawler access, put real buyer questions to AI tools, map which sources the answers cite, and fix the errors you find. No tool can guarantee you a place in an answer. There are five steps. Check that AI crawlers and search indexes can reach your pages, write a set of real buyer questions, put them to each AI tool several times, then map the sources the answers cite and check every fact stated about you. "AI SEO audit" and "GEO audit" are other names for the same job, and the checks stay the same whatever you call it. Each point below is tagged. **Documented** means a platform or standard publishes it. **Research** means a study reported it under its own test conditions. **Recommendation** means it is our advice on sensible practice, with no official backing. For background on the term itself, see [what AI visibility is](/what-is-ai-visibility/). ## What an AI visibility audit is, and what it is not An audit is a snapshot: a structured look, on particular dates, at whether AI systems can read your site and what they say when buyers ask about your field. What you get is evidence: where you appear, where you do not, what is said about you, and which pages the answers lean on. It is no score and no promise. None of the platform documents listed on this page promises inclusion in an AI answer or gives a timeframe, so an audit that implies it can get you into ChatGPT is overselling. Some vendors publish audit scores and averages of their own. Those numbers belong to the vendor and depend on the vendor's own method, so we do not repeat them here. ## How it differs from a classic SEO audit and from ongoing measurement A classic SEO audit asks whether your pages can rank in a list of links. This audit asks whether you get named, cited and described correctly inside a written answer. Much of the groundwork overlaps. For Google, the overlap is close to total. Google says that to be eligible as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to show in Google Search with a snippet, and that there are no additional technical requirements (documented). Bing says its Bing and Copilot search experiences rely on the same crawling, indexing and ranking foundation as traditional search (documented). So an indexing problem found in an SEO audit is an AI visibility problem too. Measurement is the other neighbour. An audit runs once, with a start and an end. Tracking the same questions on a schedule is a separate job, covered in the [measurement section](/guide/#measurement) of our main guide. ## Step 1: Check access (robots.txt, blocking, rendering, indexing) Start here, because the later steps depend on it. Open `yourdomain.com/robots.txt` and read it line by line, looking for blanket `Disallow` rules and for rules aimed at AI user agents that someone added and forgot about. The agents do different jobs and are controlled separately (documented, from each vendor's crawler page): | Platform | Agent | What the vendor says it does | robots.txt note | |---|---|---|---| | Google | Googlebot | Crawls for Search, including AI Overviews and AI Mode | Google says Googlebot rules are the control for Search | | OpenAI | OAI-SearchBot | Surfaces sites in ChatGPT search results | Opted-out sites are not shown in ChatGPT search answers, though they can still appear as navigational links; OpenAI says changes can take about 24 hours to take effect | | OpenAI | GPTBot | Crawls content that may be used to train models | Disallowing it signals content should not be used for training | | OpenAI | ChatGPT-User | Visits pages for certain user actions | Robots.txt rules may not apply, because a user started the action | | Perplexity | PerplexityBot | Surfaces and links sites in Perplexity search results | Perplexity recommends allowing it if you want to appear in its results | | Perplexity | Perplexity-User | Fetches pages when a user asks a question | Generally ignores robots.txt, because a user requested the fetch | | Anthropic | ClaudeBot | Collects content that could contribute to training | Anthropic says its bots honour robots.txt | | Anthropic | Claude-User | Fetches pages when people ask Claude questions | Blocking it may reduce visibility for user-directed web search | | Anthropic | Claude-SearchBot | Indexes content to improve search results | Blocking it may reduce visibility in search results | For each row, write down allowed or blocked, and whether that was a deliberate decision (recommendation). Blocking a training crawler and blocking a search crawler are different choices with different effects. Keep in mind what robots.txt is. RFC 9309, the standard, says its rules "are not a form of access authorization" and that the protocol is no substitute for real security measures (documented). It asks crawlers to stay out; it cannot stop anything. Then check four more things: 1. **Firewalls and CDNs.** Google's AI features page lists making sure crawling is allowed "by any CDN or hosting infrastructure" as well as in robots.txt, and Perplexity's crawler page explains how to allow its bots through a web application firewall (documented). Look at your bot-protection settings as well as robots.txt. 2. **Rendering.** Bing advises against hiding critical content behind client-side rendering and says content that cannot be reliably rendered may not be indexed or selected for grounding; Google advises making important content available in textual form (documented). Fetch the raw HTML of your key pages and check the main text is there. 3. **Indexing.** Confirm your important pages are indexed in Google Search Console and Bing Webmaster Tools (recommendation, building on the documented indexing requirement above). 4. **Meta directives.** Google says `nosnippet`, `data-nosnippet`, `max-snippet` and `noindex` limit what is shown from your pages in Search. Bing says `NOARCHIVE` stops content being used in Copilot responses and `NOCACHE` limits Copilot to the URL, title and snippet (documented). Make sure any of these on your key pages are there on purpose. More on crawler rules is in the [crawler access section](/guide/#crawler-access). ## Step 2: Build a buyer prompt set Build the questions from what real buyers say (recommendation): sales calls, support emails, enquiry forms, review sites and the searches that already bring people to you. Prompts made up in a meeting mostly test your own assumptions about what your market asks. Group them by intent: - **Category questions**, where the buyer has not named anyone yet ("who does X near Y", "how do I choose a Z"). - **Comparison questions**, where they are weighing options. - **Brand questions**, where they name you directly ("is [your business] any good"). Include all three. Showing up when someone types your name says little about whether you show up when a stranger asks a general question. Write each prompt the way a person would type it, and freeze the wording so later runs can be compared. ## Step 3: Run the prompts and record what comes back Put every prompt to every platform you care about, and run each one more than once (recommendation). Answers vary, so one run proves little. Google's own documentation says AI Overviews and AI Mode may use different models and techniques, so the responses and links they show will vary, and that AI Overviews often do not appear at all (documented). The GEO researchers averaged their experiments over five random seeds to reduce variance (research). That was their lab setup and sets no target for you; it does show why repeat runs matter. Log the conditions for every run (recommendation): date, platform, logged in or not, location, and the model or mode used. Answers can differ with these, and without the log you cannot tell a real change from a different setup. For each answer, record in a spreadsheet (recommendation): | Column | What to write | |---|---| | Named | Yes or no | | Position | Where in the answer you appear, and who is named before you | | Sentiment | Positive, neutral, negative or mixed, with the words used | | Cited URLs | Every link given, yours or anyone else's | | Errors | Any wrong fact about you, quoted exactly | Paste in the full answer as well. Summaries written from memory drift. ## Step 4: Map the sources the answers cite Sort the cited URLs by domain. Which keep turning up: your own pages, competitors, review sites, forums, directories, news outlets? This layer often shows the most (recommendation). If answers in your category keep citing three comparison pages and you are on none of them, you have found something no on-site check would show. If they cite your page for one topic and a competitor's for another, that is a content gap. Tag each source as yours, a third-party page that mentions you, or a page with no link to you at all. Do not over-read one citation: a URL that appears once is an anecdote, while one that appears across runs and platforms is a pattern. If Google matters to you, look at Search Console as well. Google says traffic from AI Overviews and AI Mode is included in the overall search traffic in the Performance report, under the "Web" search type (documented). ## Step 5: Check how AI describes your business (accuracy and entity facts) Read every description against the facts. Is your name right? What you sell, where you work, who you serve, your prices if mentioned, whether you are still trading? Mark each mismatch, then trace it (recommendation). Errors often come from an old page, a stale directory listing or an outdated article, so check the sources from Step 4. A wrong fact and the page feeding it are often the same finding. Google's AI features page also advises keeping your Merchant Center and Business Profile information up to date (documented). Record vague answers as vague rather than wrong. The difference matters when you prioritise. ## How to read the results without fooling yourself **Treat variation as normal.** A mention in one run and none in the next is no contradiction; it is the thing you are measuring. Report how often you were named across runs. **Distrust single numbers.** A score hides the spread you worked to capture. If you see a vendor's score or benchmark quoted, it is that vendor's figure and only comparable with other figures from the same tool. **Keep the research in proportion.** Aggarwal et al., "GEO: Generative Engine Optimization" (KDD 2024), tested content changes such as citing sources, adding quotations and adding statistics on a 10,000-query benchmark called GEO-bench. Their generative engine was a language model answering from the top five Google results, scored with the paper's own visibility metrics. They report gains of up to 40% on those metrics, found results varied by domain, and also ran tests on Perplexity (research). That is a useful source of ideas for fixes. It does not show what will happen to your site on a live platform today. ## Turning findings into a prioritised fix list Rank problems by how much they block and how hard they are to fix (recommendation): 1. **Access problems.** A blocked search crawler, a firewall rule or an unindexed key page stops everything downstream. Fix these first. 2. **Factual errors.** Wrong claims about your business do active harm, and the page feeding them is often one you or a directory can correct. 3. **Source gaps.** Absence from the third-party pages that answers keep citing. These take longer, because they sit outside your site. 4. **Content gaps.** Buyer questions where you are not mentioned and have no page that answers them. Give each line an owner and a rough effort. Fixes make you easier to find and describe correctly; they do not guarantee a mention or a date. For platform-specific detail, see [how to rank in ChatGPT](/how-to-rank-in-chatgpt/) and [how to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/). ## Doing it yourself, with tools, or hiring someone Doing it by hand costs time and teaches you the most: reading many answers about your own business shows things a dashboard summary hides. Tools save effort when you have many prompts or several platforms, because they automate the running and logging. Judge them on whether they show you the raw answers and cited URLs as well as any summary number. Hiring someone makes sense when you lack time or want an outside view; apply the same test and ask to see the underlying answers. Whichever route you take, the method above is the same. The [how AI answers happen](/guide/#how-it-works) section shows where each step fits. ## When to repeat the audit Repeat it after meaningful changes (recommendation): a site migration, a robots.txt or firewall change, a rebrand, a big content push, or a fix to the errors you found. Those are the moments the snapshot goes out of date. If you want a regular rhythm with the same prompts, you have moved from auditing into tracking. The [measurement section](/guide/#measurement) covers that. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you would rather not rerun the prompt set by hand, it tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions. - [Hilyt](https://hilyt.it/): If Step 5 turns up wrong facts about your business, it gives you an open, AI-readable profile that sets out your details in one place. ## Questions people ask ### What is an AI visibility audit? It is a one-off check of whether AI crawlers and search indexes can reach your site, what AI answers say about you for real buyer questions, and which sources those answers cite. It is a snapshot taken on particular dates, done by hand or with a tool. ### How do I run an AI visibility audit for free? Check robots.txt, firewall rules and indexing, then write prompts from real buyer questions and put each one to the AI tools you care about more than once. In a spreadsheet, record whether you were named, where, the tone, the cited URLs and any errors, plus the date, platform, login state, location and mode. ### What is the difference between an AI SEO audit, a GEO audit and an AI visibility audit? Mostly the label. All three check access, answers and sources for AI-generated responses. "GEO" comes from generative engine optimisation, the term Aggarwal et al. introduced in their KDD 2024 paper. ### How many prompts should I test, and how often should I repeat them? No platform or study we read gives a number for a business audit, so we do not give one. Take prompts from real buyer questions across category, comparison and brand intent, run each more than once, and repeat the audit after significant changes such as a migration or a robots.txt edit. ### Can an audit guarantee my site will be mentioned by ChatGPT? No. None of the platform documents we read promises inclusion or a timeframe. An audit shows a snapshot and finds fixable problems, such as blocking OAI-SearchBot, which OpenAI says keeps a site out of ChatGPT search answers. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Bing Webmaster Guidelines](https://www.bing.com/webmasters/help/webmaster-guidelines-30fba23a) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full text (HTML)](https://arxiv.org/html/2311.09735v3) --- # Guide: AI visibility checker: what a free check tells you, and how to run your own Canonical: https://whatisgeo.app/ai-visibility-checker/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > An AI visibility checker asks AI assistants questions or scans your site, then reports whether you are named or cited. Treat one run as a sample, not a score, and check by hand too. Whatever a checker reports, whether a percentage, a grade or a list of mentions, comes from a small number of answers collected on one day for the questions the tool chose. That makes it a useful snapshot and a poor verdict. This page explains the three kinds of checker, what each can and cannot show, and how to run your own check for free. ## What an AI visibility checker actually does Under the dashboard, a checker does one of two jobs, and some tools do both. It puts questions to AI assistants and records the replies. Or it looks at your website and reports how ready it seems to be for assistants to find and use. Neither job sees how every user is being answered. The tool has the answers it collected, on the day it collected them, for its own prompts. Keep that in mind and the output becomes much easier to read. If you want the definition of visibility itself first, see [what is AI visibility](/what-is-ai-visibility/). ## Three kinds of checker and what each one measures The three labels below are ours, for clarity. They are not an industry standard, and one product can mix them. | Kind | What it does | What it can show | What it cannot show | | --- | --- | --- | --- | | Answer sampling | Sends prompts to one or more assistants and reads the replies | Whether you were named or linked, and who else appeared, in those runs | Whether that holds across other runs, phrasings or days | | Site readiness scan | Crawls your pages and checks access and structure | Whether crawlers are blocked, whether content is readable as text | Whether any assistant has actually mentioned you | | Simulated answers | Uses a model to predict how an assistant might answer, often against a rival's content | Possible gaps in your content | What a real assistant says to a real person | The three answer different questions. Sampling asks "was I named?" A readiness scan asks "could I be?" A simulation asks "might I be?" Most misreadings of checker results come from treating one of these as if it answered another. A readiness scan is on firmest ground when it checks things platforms document themselves. Google's guidance on AI features lists, among its SEO basics, making sure crawling is allowed in robots.txt and by any CDN or hosting setup, and making sure important content is available as text. Anything a scanner scores beyond documented requirements like these is the vendor's own judgement, so read it as their judgement. ## What a single free check cannot tell you A free check is usually one run. Answers from AI assistants can differ when you ask the same question again, so our recommendation is to treat any single run as one draw from a range of possible answers. That is a working rule from practice, not a measured statistic. If the same prompt gives different brands on two tries, the assistant is behaving normally and the checker has not broken. So one run leaves three questions open. 1. **Consistency.** You might be named once and missing the next few times, or the other way round. 2. **Phrasing.** Buyers ask the same thing many different ways. A checker only tests the wording it was given. 3. **Readiness versus outcome.** Good access is good news, and it is only a starting point. For Google, a page must be indexed and eligible to appear with a snippet to be shown as a supporting link in AI Overviews or AI Mode, and Google says there are no additional technical requirements. Eligibility is all that gives you. It does not promise inclusion. Google also notes that AI Overviews and AI Mode may use different models and techniques, so the responses and links they show will vary between the two. A check of one AI feature tells you little about another. ## How to run your own check in 15 minutes You need a spreadsheet and two or three assistants. The time below is our estimate for a first pass with five prompts; more prompts or assistants take longer. 1. **Write five prompts.** Use the questions a real customer would type: two or three about your category ("best accountant for freelancers in Leeds"), one comparing options, one naming your business directly. Keep the wording plain. 2. **Pick two or three assistants** your customers are likely to use, and keep the same set every time you repeat the check. 3. **Use a fresh chat for each prompt** so earlier answers do not shape later ones. 4. **Log each answer.** Note whether your business was named, whether your site was linked as a source, which competitors appeared, and anything it got wrong about you. Yes or no per column is enough. 5. **Check your robots.txt.** Make sure you have not blocked the crawlers that matter by accident. OpenAI, for example, documents separate agents: OAI-SearchBot is used to show sites in ChatGPT's search features, and OpenAI says sites that opt it out will not appear in ChatGPT search answers, though they can still appear as navigational links; GPTBot crawls content that may be used for training; ChatGPT-User visits pages when a user asks something, and OpenAI says robots.txt rules may not apply to it. The rules work under RFC 9309, which describes them as rules crawlers are requested to honour and says they are not a form of access authorisation. For a full walkthrough, see [AI crawlers and robots.txt](/ai-crawlers-robots-txt/). 6. **Date the sheet.** A check with no date cannot be compared with the next one. Then run the same prompts again on another day. Two dated passes tell you more than one long session. ## How to read the result and what to do next Look for patterns across runs and assistants, not single answers. If you are named when someone asks about you by name but rarely for category questions, people who already know you can find you and new buyers mostly cannot. If you never appear at all, rule out access first (step 5), then look at whether your pages answer the questions you asked. Do not read much into one missing mention. A pattern that repeats across prompts, assistants and days is worth acting on. Fix access problems first, because they are usually cheap and clear. If you want a structured review, an [AI visibility audit](/ai-visibility-audit/) works through access, prompts, cited sources and accuracy in order. For a repeatable method with scoring, see [how to measure AI visibility](/how-to-measure-ai-visibility/), and for the other signals worth tracking, including Search Console, which Google says counts AI Overviews and AI Mode traffic in its Performance report under the Web search type, see [measuring progress](/guide/#measurement). Repeat the manual check on a regular schedule. How often is up to you; we suggest monthly as a sensible default for a small site. The trend across dated checks matters more than any single result. ## What to look for in a checker, free or paid We do not rank checkers, because the right one depends on what you need to know. These questions apply to any of them: - **Does it say which kind it is?** It should be clear whether it sampled real answers, scanned your site or simulated a reply. If that is vague, be wary. - **Can you see the raw answers?** A score with no evidence behind it is hard to trust. You want the actual replies so you can judge them yourself. - **Does it run each prompt more than once?** Because answers vary, repeated runs say more than a single snapshot. - **Does it say which assistants, settings and dates it used?** Without that you cannot compare one check with the next. - **Do its claims come with sources?** Any claim about what a tool does, or about its accuracy, should come from the vendor's own pages. Treat it as the vendor's claim unless someone independent has tested it. For a side-by-side view of the tool categories, see [generative engine optimisation tools](/generative-engine-optimization-tools/). You may not need a tool at all. For a small business, the manual check above may be all you ever need. A bigger site with many products or markets is where a tool's value shows: it saves you repeating the same prompts by hand. Paying for that is reasonable. Paying for a score you cannot inspect is not. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If rerunning the manual check by hand gets tedious, it runs buyer questions across ChatGPT, Claude, Gemini and Perplexity and records whether your company is named and cited; ask it the same questions you would ask any checker. ## Questions people ask ### What is an AI visibility checker? It is a tool that puts questions to AI assistants, scans your site, or does both, and then reports whether your business is named or cited. What it shows you is a small set of observations from one day, so it is not a full picture. ### Are free AI visibility checkers accurate? A free check can be accurate about the answers it collected and still mislead you, because one run is a single observation and answers can change between runs. We know of no independent accuracy figure for these tools. Repeat the check and compare it with your own manual results. ### Can I check my AI visibility without a tool? Yes. Write five real buyer questions, ask them in two or three assistants in fresh chats, and note whether you are named and whether your site is linked. Check your robots.txt, put a date on the sheet and repeat it later. ### What is the difference between an AI visibility checker and a tracker? A checker is usually a one-off look. A tracker runs the same questions again on a schedule so you can see movement over time. Since any single run is a sample, the repetition is where most of the value comes from. ### Does a high readiness score mean AI will cite me? No. A readiness score is about access and structure, and it cannot tell you what an assistant chooses to say. Google, for example, says being indexed and eligible for a snippet makes a page eligible for AI Overviews and AI Mode, and it promises nothing beyond eligibility. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) --- # Guide: AI-driven SEO services: what you are buying, what Google says, and how to vet one Canonical: https://whatisgeo.app/ai-driven-seo-services/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-26 > "AI-driven SEO services" can mean SEO done with AI tools or SEO aimed at AI answers. Ask which you are buying, who checks the work and how results are measured. Nobody can guarantee rankings or AI citations. Before you compare providers, find out which of the two meanings a proposal uses, because the deliverables, the risks and the way you measure them all differ. This guide sets out what Google documents, what research shows and what is just sales copy, then gives you questions to ask and red flags to watch for. ## Two things the phrase can mean The first meaning is about **method**. The provider uses AI tools to do the work: keyword clustering, content briefs and drafts, audits, internal linking suggestions. What you get is still ordinary SEO, done with different tools. The second meaning is about **target**. Here the service tries to get your business named, described or cited in generated answers, such as a ChatGPT reply or a Google AI Overview. That is a different goal and needs different measurement. For more on the second meaning, see [what is AI SEO](/what-is-ai-seo/) and our guide to hiring a [generative engine optimisation agency](/generative-engine-optimization-agency/). Providers often blur the two. A proposal can say "AI-driven" and mean either, both, or neither once you read the scope. So the first job is to make them say which. ## What AI-driven SEO services typically include There is no standard scope. What follows are common patterns from our reading of this market, so treat it as a description and not a definition. Expect some mix of: 1. Technical and content audits, often run with tools. 2. Keyword and topic research, often clustered with AI. 3. Content briefs or full drafts written with language models. 4. On-page changes: titles, headings, internal links, structured data. 5. Reporting. For the second meaning, this may include checks on how AI answers mention your brand. Some providers chain these tasks into automated workflows. That can save time, but it makes review more important, because errors get produced faster and in bulk. ## What Google documents about AI-generated content and scaled pages This is the one part of the subject where there are written rules, so it is worth reading them in Google's own words. **AI use by itself is allowed.** Google's 2023 guidance on AI-generated content says "appropriate use of AI or automation is not against our guidelines", and that its focus is on "the quality of content, rather than how content is produced". It also says using AI "doesn't give content any special gains". **Scaled content abuse is not allowed, however it is made.** Google's spam policies define scaled content abuse as many pages "generated for the primary purpose of manipulating search rankings and not helping users", producing little or no value "no matter how it's created". The first example on the list is "using generative AI tools or other similar tools to generate many pages without adding value for users". Google says sites that violate its spam policies "may rank lower in results or not appear in results at all". The same page also counts attempts to manipulate generative AI responses in Google Search as spam. **Review is expected.** Google's guidance on generative AI content says it is "critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing", and that this applies to titles, meta descriptions, structured data and image alt text as well. **AI Overviews and AI Mode have no extra requirements.** Google's page on AI features says a page must be indexed and eligible to show in Search with a snippet, and "there are no additional technical requirements". It adds that "specific optimization isn't required" and that normal SEO fundamentals still apply. How this maps onto a proposal: | What the service proposes | What Google's documentation implies | |---|---| | AI-assisted drafts, edited and fact-checked by a person | Allowed; Google judges the quality of the result | | Hundreds of templated pages that differ mainly by keyword | Matches the scaled content abuse pattern if the pages add little for users | | AI-written titles, descriptions and schema pushed live unreviewed | Google says to review these before publishing | | A special technique for getting into AI Overviews or AI Mode | Google documents no extra technical requirements | We cover the content question in more depth in [is AI content good for SEO](/is-ai-content-good-for-seo/). ## What is research and what is sales claim Very little in this market is established fact, so be strict. There is real research. Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024, built a benchmark called GEO-bench, made up of user queries across many domains along with web sources to answer them. In that benchmark, the authors report that their methods could raise visibility in generative engine responses by up to 40%. They also report that how well the methods worked varied by domain. That is a result under controlled test conditions. It does not tell you what will happen to your site, in your sector, on a live product that changes over time. Treat everything else as a claim until you see the evidence. That includes case studies with no method, dashboards with no definitions, and "proprietary" models offered in place of evidence. Some of these claims may be true, but if you cannot check a claim, it is no reason to sign. ## Questions to ask before you sign These are recommendations, not documented requirements. 1. Which meaning of "AI-driven" do you mean: AI-assisted delivery, visibility in AI answers, or both? 2. Which parts of the work do people do, and who reviews and fact-checks anything published under my name? 3. How do you make sure every page you publish adds something for a reader that the others do not? 4. What will you measure, how, and how often? Can I see the definitions and keep my own copy of the data? 5. What access do you need, and what can you change without asking me? 6. Which tools and models do you use, and does any of my content or data go into them? 7. What happens to the content and the data if we stop working together? A good provider can answer all of these plainly. ## Red flags in proposals Again, these are recommendations, though some tie back to Google's documentation. | Red flag | Why it matters | |---|---| | A guarantee of rankings, traffic or AI citations | Nobody outside the platforms controls the results | | Large-scale page production as the main deliverable | Can match Google's scaled content abuse pattern | | No named person reviewing content before it goes live | Google says AI output should be fact-checked by hand before publishing | | A special technique for Google's AI features | Google documents no extra technical requirements beyond normal Search eligibility | | Vague scope, such as "AI optimisation", with no list of deliverables | You cannot check what you paid for | | Results shown without method, baseline or dates | You cannot tell the provider's effect from normal variation | | Refusal to name the tools or models used | You cannot judge data handling or quality control | One of these may just mean a badly written proposal. Several together suggest a provider to avoid. ## How to judge whether it is working Agree on the measures before work starts. For standard SEO, that means search visibility, clicks, and the pages that bring the outcomes you care about. Google says traffic from AI Overviews and AI Mode is included in the Search Console Performance report under the Web search type, so no separate AI figure exists for a provider to point to. Our home guide covers [measurement](/guide/#measurement) in more detail. For other AI answers, it is harder. Answers vary between runs and between platforms, so one screenshot proves very little. As a sensible practice, track a fixed set of buyer questions over time. Note whether your business is named and whether it is described accurately. Compare against a baseline you recorded before the work began. Our guide on [how to measure AI visibility](/how-to-measure-ai-visibility/) walks through the method. Ask for a change log too: what changed, when and why. Without one, you cannot link any movement to the work. If the proposal covers crawler access, check each platform's own documentation and avoid relying on the provider's summary. Our guide to [AI crawlers and robots.txt](/ai-crawlers-robots-txt/) links to those documents. ## When you may not need a service at all Sometimes the honest answer is that you do not need one. Google says the same SEO fundamentals apply to its AI features: crawling allowed in robots.txt, content findable through internal links, important content in text form, and structured data that matches what is on the page. If your site is small, your offer is clear and someone on the team can write well, doing these basics yourself may go further than a retainer. A service is worth paying for when you lack the time, skills or capacity and can name the specific gap. If you cannot name the gap, you may be paying for reassurance more than for work. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want your own baseline to judge a provider against, it tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions. ## Questions people ask ### What are AI-driven SEO services? Providers use the label for two different things. One is ordinary SEO delivered with AI tools. The other is work aimed at getting a business named or cited in AI-generated answers. Ask which one a proposal means and get the deliverables in writing. ### Is an AI-driven SEO service different from a generative engine optimisation agency? Often, though the two overlap. A generative engine optimisation agency focuses on visibility in AI answers, while an AI-driven SEO service may just use AI to do standard SEO. Compare the written scope of work and ignore the label. ### Does Google penalise content produced with AI? Not for that reason alone. Google says appropriate use of AI or automation is not against its guidelines. Using AI to generate many pages without adding value for users can break its scaled content abuse policy, and sites that break its spam policies may rank lower or not appear at all. ### How much do AI SEO services cost? We do not quote prices because they depend on scope and provider, and we have not checked vendors' own pricing pages for this guide. Ask each shortlisted provider for a written scope and fee so you can compare like for like. ### Can an AI SEO service guarantee I appear in ChatGPT or AI Overviews? No. Nobody outside the platforms controls what an AI answer says, and Google says there are no extra technical requirements for its AI features beyond being indexed and eligible to show with a snippet. Treat any guarantee as a red flag. ## Sources - [Google Search Central: Spam policies for Google web search](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: Guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: Are Google AI Overviews accurate? What the studies show, and what to do about errors Canonical: https://whatisgeo.app/are-google-ai-overviews-accurate/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Often, but not reliably. Google says AI Overviews can and will make mistakes, and the main outside test found right answers about 91% of the time on hard factual questions, with weaker backing from cited sources. It depends on the question, and on whether an answer still counts as accurate when the pages it cites do not back it up. In the one large outside test we could trace to its original write-up, only 39% of overviews were both correct and fully supported by their cited sources. In practice, treat an overview as a lead and check anything that matters. This page keeps three kinds of claim apart: what Google documents, what a named analysis measured (and its limits), and what we recommend. If you want the basics first, start with [what Google AI Overviews are](/what-are-google-ai-overviews/). ## What Google says about accuracy and mistakes **Google's documentation.** The Search Help page on AI Overviews says generative AI "may provide inaccurate or offensive information" and that "AI Overviews can and will make mistakes". It also carries a shorter warning: "AI responses may include mistakes." Google gives no accuracy percentage on that page or on its Search Central page "AI features and your website". **Google's account of how overviews are built.** In "AI Overviews: About last week" (30 May 2024), Google's VP of Search, Elizabeth Reid, wrote that the model behind AI Overviews is "integrated with our core web ranking systems" and is "built to only show information that is backed up by top web results". She said overviews "generally don't 'hallucinate'" in the way other chatbots can. When they go wrong, she wrote, it is usually from "misinterpreting queries, misinterpreting a nuance of language on the web, or not having a lot of great information available". The post also says Google's own tests put AI Overviews' accuracy "on par with" featured snippets. That is Google's claim about its own product, and the post gives no figure or method. **What Google changed after the May 2024 errors.** The same post admits that "some odd, inaccurate or unhelpful AI Overviews certainly did show up", alongside many faked screenshots. Examples included satire about eating rocks and a forum joke about putting glue on pizza. Google listed fixes, including: - better detection of nonsensical queries, and limits on satire and humour content - limits on user-generated content "in responses that could offer misleading advice" - triggering restrictions for queries where overviews were not proving helpful - an aim not to show overviews for hard news topics, and extra triggering refinements for health ## What the outside analysis found, and its limits The figure most often repeated, roughly 91 percent correct, comes from Oumi, an AI model company that ran the test at the request of The New York Times. Oumi published its method on 14 April 2026. | Item | What Oumi reported | |---|---| | Questions | SimpleQA, a set of a little over 4,000 short, hard factual questions with known answers, built by OpenAI | | How overviews were collected | Automated Google searches, a screenshot of each overview, text extracted by OCR, cited pages scraped | | How correctness was judged | An AI model (GPT-5) compared each overview with SimpleQA's answer | | How source support was judged | Oumi's own unreleased model, HallOumi++, checked every claim against the cited text | | Human checking | People hand-checked the judging models on a small sample (the post gives 25 overviews as an example) | | Sample assessed | 1,609 queries with overviews powered by Gemini 3 (early 2026) | | Correct answer | About 91% | | Correct, with every claim supported by cited sources | 39% | | Claims supported, across 4,876 claims | 67% | Oumi also compared a late 2025 run (Gemini 2) with the early 2026 run (Gemini 3) on the 158 queries both could assess. Correct answers rose from 89% to 94%, while fully supported answers fell from 52% to 38%. **The limits matter as much as the numbers:** - **The questions are unusual.** SimpleQA is a research benchmark of obscure facts. Ars Technica reports that a Google spokesperson told the Times the study "has serious holes" and "doesn't reflect what people are actually searching on Google", and that Google believes SimpleQA contains incorrect information. Oumi replied that Gemini models score similarly on SimpleQA Verified, a vetted version Google prefers. - **"Correct" is narrow.** Oumi says correctness covers only the question asked, not the other facts in the overview. - **"Unsupported" may overstate the problem.** Oumi says it cannot know everything Gemini is given. Overviews may draw on text outside the quoted fragments, or on sources they do not show. Only queries whose citation fragments could all be located were analysed. - **AI judged AI.** Both judges are language models, with human checks on a small sample. - **The analyser has a product to sell.** HallOumi is Oumi's own technology, and the post promotes Oumi's platform. Some headlines turn the 9 percent miss rate into millions of wrong answers per hour or per year. Those are extrapolations by journalists and commentators, built on their own assumptions about search volume and how often overviews appear. Oumi's post does not calculate one, so we have left those figures out. ## Why AI Overviews get things wrong Google's explanation, quoted above, names three causes: misreading the query, misreading the nuance of a web page, and "data voids" where little good information exists. Its rock-eating example was a data void, filled by satire that a real company had republished on its own website. Beyond that, what follows is our general explanation rather than a documented finding. An overview is generated text that summarises retrieved pages. When sources disagree, a summary can blend them or pick one without saying so. When a sentence loses its caveats, its meaning can shift. And an answer can be right and still cite pages that do not quite say it, which is the gap Oumi was measuring. For how retrieval and generation fit together, see [how AI answers happen](/guide/#how-it-works). ## How to check an AI Overview before you rely on it Google's help page gives three tips: check important information in more than one place by clicking the supporting links and trying other results; ask your question in different ways; and send feedback if something looks wrong. We would add four recommendations of our own: 1. **Open the cited links** and find the sentence that supports the claim. If you cannot find it, treat the claim as unconfirmed. 2. **Compare two independent sources**, ideally including a primary one such as an official body or the company itself. 3. **Check the dates** on the sources. An answer can be faithful to a page that is out of date. 4. **Take extra care with health, legal and money questions.** Use the overview to find sources, then rely on the sources or a qualified professional. Google's help page says AI Overviews cannot be turned off, but the Web filter on a results page shows only text links without them. ## How to report a wrong AI Overview Google's Search Help page describes this route: 1. At the bottom of the overview, select the thumbs down icon if you found it unhelpful, inaccurate, biased or otherwise problematic. 2. Select **Report a problem** and choose the category that best describes the issue. 3. Add details in the text box. Google recommends this. 4. Select **Submit**. Google says the feedback includes your most recent search and its results. The page adds that you can also give feedback from the three-dot menu at the top right of an overview. In the details, say what the overview claimed, what is correct and where the correct version is published. Nobody can guarantee that a report will change the overview, or say when it might. ## What to do if an AI Overview gets your business wrong Start by capturing the evidence: the exact search, a screenshot, the date and the cited links. The cited pages often show where the wrong fact came from. Then work through these steps. The first two come from Google's Search Central guidance on AI features; the rest are our recommendations. None of them guarantees a correction or a timeframe. 1. **Keep your Business Profile and Merchant Center details up to date.** Google lists this among its SEO best practices for AI features, along with making sure structured data matches the visible text on the page. 2. **Make important facts available as text on your own pages**, which is also on Google's list. State prices, locations, services and opening hours plainly, in one place, and keep them current. 3. **Fix contradictions elsewhere.** If an old directory listing, press release or review site carries the wrong detail, correct it at the source where you can. 4. **Report the overview** using the steps above, and link to the page with the correct information. 5. **Check again over time.** An [AI visibility audit](/ai-visibility-audit/) sets out how to record what AI tools say about you, and [how to track AI Overviews](/how-to-track-ai-overviews/) covers monitoring the overviews themselves. For measuring progress in general, see [measuring progress](/guide/#measurement). If your concern is being cited at all, see [how to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want to know how AI assistants describe your business, it checks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for the questions your buyers ask. ## Questions people ask ### Are Google AI Overviews accurate? Often, but not reliably. Oumi's test for The New York Times found about 91% of overviews gave the right answer to hard factual questions, but only 39% were correct and fully supported by their cited sources. Google itself says AI Overviews can and will make mistakes. ### Why is the Google AI Overview wrong? Google says errors usually come from misreading the query, misreading the nuance of a web page, or a lack of good information on the topic. It also limited satire and some user-generated content after odd answers in May 2024. ### How do I report a wrong or harmful AI Overview? Google's Search Help page says to select the thumbs down icon at the bottom of the overview, then Report a problem, choose a category, add details and submit. You can also give feedback from the three-dot menu on the overview. Nobody can guarantee the overview will change. ### Can I trust AI Overviews for health, legal or money questions? We recommend using them only to find sources. Google's own advice is to check important information in more than one place, so open the cited pages and rely on primary sources or a qualified professional. ### Can I correct what an AI Overview says about my business? You cannot edit it directly. You can report it, keep your Business Profile and website facts accurate and consistent, and fix wrong details on other sites. None of these guarantees a correction or a timeframe. ## Sources - [Google Search Help: AI Overviews in Google Search](https://support.google.com/websearch/answer/14901683) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google blog, 30 May 2024: AI Overviews: About last week](https://blog.google/products/search/ai-overviews-update-may-2024/) - [Oumi, 14 April 2026: Oumi's Study Finds 50% of AI Overviews Untrustworthy (method and findings)](https://oumi.ai/blog/oumis-study-finds-50-of-ai-overviews) - [Ars Technica, April 2026: Testing suggests Google's AI Overviews tell millions of lies per hour (includes Google's response)](https://arstechnica.com/google/2026/04/analysis-finds-google-ai-overviews-is-wrong-10-percent-of-the-time/) --- # Guide: Are there ads in Google AI Overviews? What Google documents, and what it means for you Canonical: https://whatisgeo.app/ads-in-ai-overviews/ Publisher: Adapt Progress Evolve Limited Updated: 2026-11-10 > Yes. Google shows ads above and below AI Overviews, and inside them in some English-language markets, labelled as sponsored. Paying for ads is not a documented route to being cited in the AI answer. Which ads you see depends on where you search. Google says ads above and below an AI Overview can run in every market that has AI Overviews, while ads inside the overview itself are limited to English in a named list of countries. Everything below comes from Google's own Ads Help page and announcements, and we say so where Google is silent. If you want the basics of the feature first, see [what are Google AI Overviews](/what-are-google-ai-overviews/). ## Where ads can appear, according to Google Google Ads Help describes two placements: | Placement | Where Google says it runs | What Google says is eligible | |---|---|---| | Above or below the AI Overview | All 200+ markets where AI Overviews are available | Standard ads, as they appear alongside other Search results | | Within the AI Overview | English, in the US, Australia, Canada, India, Indonesia (AI Overviews only), Kenya, Malaysia, New Zealand, Nigeria, Pakistan (text ads only), the Philippines and Singapore | Text and Shopping ads from existing Search, Shopping and Performance Max campaigns | The UK was not on the list for ads within AI Overviews when we checked the Help page on 10 November 2026. The page says the list may not be exhaustive and that Google keeps expanding it, so check the page itself before you plan around it. The timeline, as Google announced it: - **3 October 2024:** Google said ads in AI Overviews were "now available for mobile users in the US" after several months of testing. - **21 May 2025:** Google said Search and Shopping ads in AI Overviews were expanding to desktop in the US, and that later that year it would expand them in English to select countries on mobile and desktop. On labelling, Google's 2024 announcement says these ads "will continue to feature our industry-leading clear and transparent 'Sponsored' label". So an ad inside an AI Overview sits in the same box as the AI-written text, but it is marked as paid. ## How an advertiser gets into AI Overviews and AI Mode There is no separate product to buy. Google says advertisers "don't need to create new campaigns or set up separate budgets", and that existing campaigns "automatically participate through the Google Ads auction when relevant". Existing Search, Shopping and Performance Max campaigns are automatically eligible. Google says it matches ads in one of two ways: directly to the user's query, when the search already expresses clear commercial intent, or contextually to both the query and the AI response, when someone asks a longer, exploratory question and commercial intent emerges from the response. Its example is someone asking why their pool is green and how to clean it, where the AI response outlines water testing, brushing and vacuuming, and Google can show ads for pool vacuum cleaners or algae treatment kits. Google says AI-powered targeting, including AI Max for Search campaigns, broad match on Search, Shopping campaigns and Performance Max, is eligible for both kinds of matching. Exact and phrase match keywords are eligible when a query expresses direct commercial intent, but are not eligible for contextual matching on broader exploratory questions. It also recommends Smart Bidding with strong conversion data and, for retailers, keeping Merchant Center product feeds complete and current. ## What advertisers can and cannot control The FAQ on Google's Help page is unusually direct: - **Targeting and opting out:** Google says "you can't target ads exclusively to AI Overviews or AI Mode, or opt out of showing ads in them separately from Search". It says negative keywords and brand exclusions still control when and how your ads serve. - **Reporting:** Google says performance metrics for ads in AI Overviews and AI Mode "are included in aggregate" in your standard Google Ads campaign reports, and the queries that trigger them are included in aggregate in the search terms report. - **Placement:** Google says an ad can appear both above an AI Overview and within it on the same search, depending on the query and auction, but not both within an AI Overview and immediately below it. What you still control is everything around the auction: which campaign types you run, your match types and bidding, your ads and creative, your product feed, your landing pages and your budget. ## Do ads change which sites the AI answer cites? Google says they do not. Its Ads Help page says "Paid advertising does not influence the organic responses generated by Google AI", and that whether your site is cited organically does not affect your ad auction eligibility. Google's Search Central documentation sets out what a page needs to appear as a supporting link in AI Overviews or AI Mode: it "must be indexed and eligible to be shown in Google Search with a snippet", and "there are no additional technical requirements". It does not mention advertising at all. A sponsored ad and a cited source are two separate things that can sit in the same box. For the organic side, [how to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/) goes into what Google does document, and the [how AI answers happen](/guide/#how-it-works) section of our main guide covers how answers get built. ## Ads in AI Mode In its May 2025 announcement Google said it was "starting to test ads in AI Mode", and that "where relevant, ads may appear below and integrated into AI Mode responses". It said advertisers already using Performance Max, Shopping and Search campaigns with broad match, including AI Max for Search, would be eligible for ads in AI Overviews and AI Mode. Google's "Ads in AI Mode" page for advertisers now says it is testing new formats for 2026, "integrated directly into AI-generated responses" while "clearly labeled" as ads, and powered by AI Max and Performance Max. Google's Ads Help page now covers AI Mode alongside AI Overviews. It says ads within AI Overviews and AI Mode are available in English in the United States and select English-speaking markets, and lists the same countries as above. Indonesia is marked as AI Overviews only, so we are not stating that AI Mode ads run there. The page says the list may not be exhaustive. If AI Mode matters to your organic work, see [Google AI Mode SEO](/google-ai-mode-seo/). ## What this means if you run ads 1. **Check what you already run.** Existing Search, Shopping and Performance Max campaigns are eligible already, and Google says you do not need new campaigns or separate budgets. Broad match, AI Max and Performance Max also cover contextual matching; exact and phrase match keywords only match direct commercial queries. 2. **Check your market.** Ads inside the overview only apply in the listed countries, in English. Elsewhere, the relevant placements are above and below. 3. **Do not promise the placement.** You cannot target it, and Google decides when it appears. Avoid telling a client or a manager that you will get them into AI Overviews. 4. **Report what you can measure.** Google reports these ads in aggregate within your standard campaign reports, so judge campaigns on clicks, conversions and cost per result. Do not estimate an AI Overview share in a report. 5. **Look after the basics.** Google's own advice for this placement is complete product feeds and good landing pages, which was already sound practice. ## What this means if you rely on organic visibility Ads and citations are separate systems, and the route to being a supporting link is the organic one Google documents: be crawlable, indexed and eligible for a snippet, and publish content that answers the question well. That is our recommendation as well as Google's guidance. It is reasonable to expect that ads in or around an overview compete for the same attention as the links inside it, but Google publishes no figures on how much. Search Console does not separate AI Overview traffic out either, so [how to track AI Overviews](/how-to-track-ai-overviews/) explains what you can and cannot measure. ## Claims about ads in AI Overviews to treat with care - **Percentages from agencies and social posts.** You will see figures about how often ads appear in AI Overviews or lose their usual position. We have not repeated any, because none we found linked to a primary source. If a number has no published method, treat it as unverified. - **Google's own performance claims.** Google says people find ads within AI Overviews helpful, citing "Google internal data, 2024". That is Google's claim about its own product, and the methodology is not published. - **"Pay to be cited" offers.** Anyone who says ad spend or a paid service will get your site cited in an AI Overview is claiming something Google does not document. Nobody can guarantee inclusion. ## Questions people ask ### Are there ads in Google AI Overviews? Yes. Google Ads Help says ads can show above, below or within AI Overviews. Ads inside the overview are limited to English in a named list of countries, and Google says they carry a "Sponsored" label. ### Can I pay to be included in an AI Overview? You can pay for an ad that may show in or around an AI Overview, but you cannot target that placement directly. Google documents no paid route into the AI-written answer or its supporting links. ### Do I need a separate campaign to show ads in AI Overviews? No. Google says advertisers "don't need to create new campaigns or set up separate budgets", and that existing Search, Shopping and Performance Max campaigns are automatically eligible. It says AI-powered targeting such as broad match or AI Max is eligible for both direct and contextual matching, while exact and phrase match keywords are eligible only for queries with direct commercial intent. ### Are there ads in Google AI Mode? Google announced tests of ads in AI Mode in May 2025 and says it is testing new ad formats in AI Mode in 2026, integrated into responses and labelled as ads. Its Ads Help page lists the same English-language markets for ads within AI Overviews and AI Mode, with some exceptions, and says the list may not be exhaustive. ### Does buying ads make my site more likely to be cited in an AI answer? No. Google's Ads Help page says paid advertising does not influence the organic responses generated by Google AI. Search Central says a page only needs to be indexed and eligible for a snippet to appear as a supporting link, with no additional technical requirements. ## Sources - [Google Ads Help: About ads in AI-powered experiences in Google Search](https://support.google.com/google-ads/answer/16297775?hl=en) - [Google Ads & Commerce blog: New ways for marketers to reach customers with AI Overviews and Lens (3 October 2024)](https://blog.google/products/ads-commerce/google-lens-ai-overviews-ads-marketers/) - [Google Ads & Commerce blog: New ways AI in Search helps your business (21 May 2025)](https://blog.google/products/ads-commerce/google-search-ai-brand-discovery/) - [Accelerate with Google: Ads in AI Mode](https://business.google.com/us/accelerate/announcements/ads-in-ai-mode/) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) --- # Guide: ChatGPT search: what it is, how it works, and how a website appears in it Canonical: https://whatisgeo.app/chatgpt-search/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > ChatGPT search is the web search feature inside ChatGPT. It pulls current web results into an answer and links the sources. Sites become eligible by allowing OAI-SearchBot, but placement is never guaranteed. For a site owner, the useful part is the citations. They are the only route by which ChatGPT sends a reader to your page. OpenAI documents one thing you control, crawler access, and says plainly that placement is not guaranteed. The launch post and help article are written for people using the feature. This page is for the people being searched: what the feature does, when it runs, where its sources come from and what you can actually change. For tactics, see the separate guide on [how to rank in ChatGPT](/how-to-rank-in-chatgpt/). ## What is ChatGPT search? OpenAI introduced ChatGPT search on 31 October 2024. At first it was available to paying Plus and Team users. Updates to the launch post say it reached all logged-in users on 16 December 2024 and everyone, without signing up, on 5 February 2025. OpenAI's help centre now lists web search on ChatGPT Free, Go, Plus, Pro, Business, Enterprise and Edu, and for signed-out users, subject to each plan's usage limits (checked 18 October 2026). Any older article that says it is for paid users only is out of date. It works on chatgpt.com and in the desktop and mobile apps. OpenAI also offers a Chrome extension that makes ChatGPT the browser's default search engine. ## When does ChatGPT search the web? There are two routes, and OpenAI documents both: - **ChatGPT decides.** The help centre says ChatGPT "may search the web automatically when your question would benefit from current information". - **The user asks.** People can pick Search from the tools menu, type / in the message box, or regenerate an answer with "Search the web". OpenAI does not publish a list of what triggers an automatic search, so treat any list you see elsewhere as guesswork. What follows for you is simple: if ChatGPT answers a question from what it already knows, no search happens and no page gets a citation, however good your site is. For how this fits into AI answers generally, see [how AI answers happen](/guide/#how-it-works). ## How does ChatGPT search work? OpenAI's help article describes the parts that matter to a site owner: 1. **It rewrites the question.** When ChatGPT search uses other search providers, it "typically rewrites your query into one or more targeted queries". After reviewing the first results, it may send further, more specific queries. 2. **It draws on outside search providers and partners.** The launch post says ChatGPT search "leverages third-party search providers, as well as content provided directly by our partners". The help article names Microsoft and Shopify among the providers whose privacy policies apply. 3. **It uses location.** ChatGPT may estimate a user's general location from their IP address and pass that general location to search providers. A question like "good restaurants near me" can become "top restaurants San Francisco". If memory is turned on, saved memories can shape the rewritten query too. 4. **It writes an answer with citations.** Responses that use search may include citations, and a Sources button lists the cited pages and other relevant links. This tells you two things. First, your page has to be findable by the searches ChatGPT runs, which partly means ordinary search indexes. Second, two people asking the same question from different places, or with different memories saved, may see different sources. One screenshot of "what ChatGPT says about us" is an anecdote. OpenAI also says it "ranks search results using multiple factors intended to help users find relevant, reliable information". It does not say what those factors are. ## Is ChatGPT search a search engine like Google? Partly. It answers questions from web pages, and OpenAI's Chrome extension can even replace Google as the browser's default search. The output is different, though: | | Google search | ChatGPT search | | --- | --- | --- | | What you get | A ranked list of results | One written answer with citations | | Where results come from | Google's own index | Third-party search providers and partner content, per OpenAI | | Crawler to allow | Googlebot | OAI-SearchBot | | Placement | Not guaranteed | "Placement is not guaranteed" (OpenAI) | With Google you compete for a place in a list. With ChatGPT search you compete to be one of the sources behind a single answer. Perplexity works in a similar way and has its own crawler; see [Perplexity SEO](/perplexity-seo/). ## Where do the sources and citations come from? From the pages ChatGPT retrieves for that conversation through the searches described above. OpenAI's own help article warns that "search results and citations can be incomplete, outdated, or incorrect" and tells users to open sources and check them. Being cited does not mean ChatGPT has described your page correctly, so read the answers you appear in. Why one page is cited and not another is where documentation ends and research begins. The best-known study is Aggarwal et al., "GEO: Generative Engine Optimization" (KDD 2024). The authors built their own generative engine, which fed the top five Google results for each query to a language model. They then tested whether rewriting a source changed how visible it was in the answers. Adding citations, quotations and statistics improved visibility in that setup, and results varied by subject. The paper also ran a smaller test on Perplexity.ai. It did not test ChatGPT search, which launched after the paper's final revision in June 2024. Treat it as evidence about generative engines in general, not a rulebook for ChatGPT. So keep three kinds of claim apart: what OpenAI documents, what a study found under its own conditions, and what practitioners recommend from experience. Most advice on this topic mixes all three in one confident tone. ## What controls does a website owner have? The documented control is crawler access. OpenAI's crawler page describes three user agents that matter here: | User agent | What OpenAI says it does | What robots.txt does | | --- | --- | --- | | OAI-SearchBot | Surfaces websites in ChatGPT's search features | Sites opted out "will not be shown in ChatGPT search answers, though can still appear as navigational links" | | GPTBot | Crawls content that may be used to train OpenAI's foundation models | Disallowing it signals your content should not be used for training | | ChatGPT-User | Visits a page when a user's request needs it; does not crawl automatically | Robots.txt rules "may not apply"; it is not used to decide what appears in search | The details OpenAI documents: - **Each setting is independent.** You can allow OAI-SearchBot to appear in search while disallowing GPTBot. - **Allow the bot and its IP addresses.** The help centre says that, to make a site eligible, you should allow OAI-SearchBot and make sure your host or CDN lets through traffic from OpenAI's published searchbot IP addresses (openai.com/searchbot.json). A firewall that blocks unfamiliar bots can undo a correct robots.txt. - **Changes are not instant.** OpenAI says it can take around 24 hours after a robots.txt update for its search systems to adjust. Allowing the crawler makes you eligible. It does not earn you a place in an answer. For the file-level rules across every AI crawler, see [AI crawlers and robots.txt](/ai-crawlers-robots-txt/), or the short version in [crawler access](/guide/#crawler-access). ## How do you check whether your site appears? There is no OpenAI console for this, so you check by hand and keep records. This is our recommended method, not an OpenAI one: 1. Write down the questions your customers actually ask. 2. Ask them in ChatGPT with Search turned on, and note whether a search ran, whether you were cited, and who was cited instead. 3. Check your server logs for OAI-SearchBot visits. Visits show the crawler can reach you. They do not show you are being cited. 4. Repeat the same questions on a regular schedule, because answers change with location, memory and time. One check is a snapshot; a series of checks shows a trend. The guide on [how to measure AI visibility](/how-to-measure-ai-visibility/) sets out a fuller method, and [what is AI visibility](/what-is-ai-visibility/) explains the term. ## What nobody can promise OpenAI's own words are "Placement is not guaranteed". No agency, consultant or tool can promise more than that. What you can do is let OAI-SearchBot in, make sure your pages are findable and worth citing, and measure honestly over time. The page-level work is covered in [how to rank in ChatGPT](/how-to-rank-in-chatgpt/). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you have checked a few questions by hand, it tracks whether ChatGPT and other assistants name and cite your company for your buyers' questions over time. ## Questions people ask ### Is ChatGPT search free? Yes. OpenAI's help centre says web search is available on ChatGPT Free, Go, Plus, Pro, Business, Enterprise and Edu, and to people who are not signed in, subject to each plan's usage limits (checked 18 October 2026). ### Does ChatGPT always search the web? No. OpenAI says ChatGPT may search automatically when a question would benefit from current information, and users can also choose Search themselves. Questions it can answer without the web may never trigger a search. ### How is ChatGPT search different from Google? Google shows a ranked list of results. ChatGPT search writes one answer and attaches citations. OpenAI also says ChatGPT search sometimes sends rewritten queries to other search providers, so it is partly built on top of other search indexes. ### How do I stop or allow ChatGPT search using my site? Use the OAI-SearchBot user agent in robots.txt. OpenAI says sites that opt out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links, and that changes can take around 24 hours to apply. ### Does blocking GPTBot keep me out of ChatGPT search? Not according to OpenAI. Its crawler page says each robots.txt setting is independent, so you can allow OAI-SearchBot for search while disallowing GPTBot, which collects content that may be used for training. ## Sources - [OpenAI: Introducing ChatGPT search (31 October 2024, with later updates)](https://openai.com/index/introducing-chatgpt-search/) - [OpenAI Help Center: Searching the web with ChatGPT](https://help.openai.com/en/articles/9237897-chatgpt-search) - [OpenAI: Overview of OpenAI Crawlers](https://platform.openai.com/docs/bots) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: ChatGPT SEO tools: using ChatGPT for SEO, and tools that track ChatGPT visibility Canonical: https://whatisgeo.app/chatgpt-seo-tools/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-20 > "ChatGPT SEO tools" means two things: using ChatGPT as an assistant for SEO work, or tools that check whether ChatGPT names and cites your site. Each has different limits, and neither can guarantee inclusion. Work out which job you need before you compare anything. If you want help writing and analysing SEO material, ChatGPT itself is the tool and its limits come from what it can see. If you want to know whether ChatGPT recommends you, you need a separate check that asks it questions and records the answers. ## What people mean by ChatGPT SEO tools One searcher wants help writing a meta description faster. Another wants to know whether ChatGPT ever names their business when a buyer asks for a recommendation. Same words, different jobs. Many results for this phrase blur the two, often because they are vendor pages making the case for their own product. That leaves you comparing a drafting assistant with a monitoring product as if they were rivals, when they do different things. Each claim on this page is marked as one of three kinds: | Label | What it means | | --- | --- | | Documented | A platform's own documentation says so (OpenAI or Google) | | Researched | A named study found it, under the conditions it describes | | Recommended | Our practical advice, which no platform has stated | For a wider look at SEO software that has added AI features, see our guide to [AI SEO tools](/ai-seo-tools/). ## Meaning one: using ChatGPT to help with SEO work Here ChatGPT is a tool you operate. You give it text, a brief or a list; it gives you text back; you decide what happens next. A useful test (recommended): could a competent junior colleague do this task working only from what you pasted into the conversation? If yes, ChatGPT can probably help. If the task needs data that colleague could not see, such as your Search Console figures, the chat alone cannot do it. ## What ChatGPT can help with, and what it cannot know Useful jobs (recommended, based on practice; OpenAI does not document these): 1. **Drafting and rewriting.** Titles, meta descriptions, outlines and first-pass paragraphs that you then edit properly. 2. **Analysing material you paste in.** Give it a page and ask where the argument is thin, or paste a list of queries and ask it to group them by intent. 3. **Structure and clarity.** Turning a messy brief into a clean outline, or flagging sections that repeat each other. 4. **Spotting gaps.** Give it two drafts and ask what one covers that the other skips. What it cannot know (recommended): your live rankings, your traffic or your Search Console data, unless you supply that data yourself. Treat anything it says about your current positions as unverified. We have not found an OpenAI page that states this directly, so read it as practical advice. On accuracy, both platforms say something on record (documented). Google's guidance on generative AI content warns that generative models predict a likely sequence of words from their training data, so their output can contain inaccuracies. It says to fact-check and review all AI-generated content before publishing, including titles, meta descriptions and structured data. OpenAI's own help article on searching the web says search results and citations can be incomplete, outdated or incorrect, and tells users to open the cited source and check it. Answers also vary (recommended): in our experience, the same question asked twice, or worded differently, can produce different answers. Check anything that matters before you act on it. Does publishing AI-assisted content hurt you with Google? Documented: Google says appropriate use of AI or automation is not against its guidelines, and that it rewards high-quality content however it is produced. Using generative AI to produce many pages without adding value for users may breach its spam policy on scaled content abuse. Our page on [whether AI content is good for SEO](/is-ai-content-good-for-seo/) covers this in more depth. ## Meaning two: tools that check whether ChatGPT names your site This is a separate job. A visibility tool does not help you write anything. It puts a set of questions to ChatGPT, records what comes back, and reports whether your business was named and whether your site was cited. Mentions and citations are different things. Documented: OpenAI's help centre says responses that use web search may include citations, which you can select to open the source. So a tool that tracks citations is in effect tracking search-backed answers. ChatGPT can also name a business without linking to it, and an outside tool cannot always tell where that mention came from (our reading; OpenAI does not say this). Our [ChatGPT search](/chatgpt-search/) guide explains how the search side works. ## What a ChatGPT visibility tool can and cannot tell you It can tell you whether your business appeared in the answers to the prompts it ran, on the days it ran them. That is a useful snapshot. Its limits: 1. **It shows what happened without the reason.** The cause is still yours to work out (recommended). 2. **It only covers the prompts it was given.** Real buyers phrase questions in ways no list fully predicts (recommended). 3. **One reading is a sample.** Because answers shift between runs and wording, look at results over time before drawing conclusions (recommended). 4. **Your customers may see something else.** Documented: OpenAI says ChatGPT may use an approximate location based on the user's IP address, and saved memories if memory is on, when it rewrites a search query. A tool's results show one view among many. 5. **It cannot guarantee inclusion.** Documented: OpenAI's help article says ChatGPT ranks search results using multiple factors and that placement is not guaranteed. No tool can change that. Our guide on [how to measure AI visibility](/how-to-measure-ai-visibility/) sets out a method, and the [measurement section](/guide/#measurement) of our main guide is the short version. What can move visibility at all? Researched: Aggarwal et al. (GEO: Generative Engine Optimization, KDD 2024) tested content changes such as adding citations, quotations and statistics. They reported visibility gains of up to 40% on GEO-bench, their benchmark of 10,000 queries, and up to 37% in a test on Perplexity.ai. The main experiments used a generative engine the authors built from the top five Google results, ChatGPT was not the engine tested, and the effects varied by domain. The study suggests what tends to help; it does not predict what a given change will do in ChatGPT today. ## What ChatGPT's own crawler and search documentation says you control This part is documented in OpenAI's crawler documentation: | User agent | What OpenAI says it is for | Effect of robots.txt | | --- | --- | --- | | OAI-SearchBot | Surfacing websites in ChatGPT's search features | Sites opted out will not be shown in ChatGPT search answers, though they can still appear as navigational links | | GPTBot | Crawling content that may be used to train OpenAI's generative AI foundation models | Disallowing it indicates your content should not be used for training | | ChatGPT-User | Visiting a page when a user asks ChatGPT or a custom GPT a question | Because the visit is started by a user, robots.txt rules may not apply; it is not used to decide whether content appears in search | The settings are independent, so you can allow OAI-SearchBot for search while disallowing GPTBot for training. OpenAI says it can take about 24 hours for search to adjust after a robots.txt change. Its help article adds that to make a site eligible for search results, you should allow OAI-SearchBot and confirm your host or content delivery network allows traffic from OpenAI's published search bot IP addresses. Recommended: check this before you buy anything. It is free, and it is the one part of the picture you fully control. ## How to choose, and a free way to check first Choose by the job (recommended): - **You want faster drafting and tidier analysis.** Use ChatGPT as an assistant, give it your own data, and have someone who knows the business review everything before it goes live. - **You want to know whether ChatGPT names you.** Run a manual check first, then decide whether a tool is worth paying for. A free manual check (recommended): 1. Confirm your robots.txt allows OAI-SearchBot and your host or CDN is not blocking it. 2. Write a handful of questions a real customer would ask before buying what you sell. 3. Ask them in ChatGPT with search switched on. OpenAI's help article says you can choose Search from the tools menu, or type / in the message box and select Search. 4. Note whether you are named, whether you are linked, and which sources are cited. Open those sources. 5. Repeat on another day and with different wording, and compare. Our [AI visibility checker](/ai-visibility-checker/) guide walks through a lightweight version. If you then look at paid tools, ask each vendor which prompts it runs, how often, from which locations, and whether you can export the raw answers. A tool that cannot show you the answers behind its score gives you little to act on. No tool decides whether ChatGPT names you. The tools only report on how your crawlable, accurate and genuinely useful pages are doing. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you need the second job done for more buyer questions than you can check by hand, it tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company; put the vendor questions on this page to it as to any other tool. - [ScriptGrain](https://scriptgrain.com/): If the first job is what you need, it drafts content in your own measured voice, and the same rule applies: someone who knows the business checks every draft before it goes live. ## Questions people ask ### Is ChatGPT an SEO tool? It is a general assistant that can help with SEO tasks such as drafting, rewriting and analysing text you give it. It has no view of your rankings or Search Console data unless you supply them, so treat anything it says about your positions as unverified (our recommendation). ### What are ChatGPT SEO tools? The phrase covers two things: ChatGPT used as an assistant for SEO work, and tools that check whether ChatGPT names and cites your site. They solve different problems, so choose by the job you need done. ### Can ChatGPT do keyword research or check my rankings? It can group keywords you give it by intent and suggest related angles. It cannot see your live rankings unless you supply the data, so check any figures it gives against your own tools (our recommendation). ### How do I check whether ChatGPT mentions my website? Ask it realistic customer questions with search switched on, and note whether you are named, whether you are linked and which sources it cites. Repeat on different days and with different wording; a visibility tool automates the same process across more prompts. ### Does Google penalise content written with ChatGPT? Google says appropriate use of AI or automation is not against its guidelines and that it rewards high-quality content however it is produced. Using AI to generate many pages without adding value for users may breach its spam policy on scaled content abuse, and Google says to fact-check AI content before publishing. ## Sources - [OpenAI: Overview of OpenAI Crawlers (OAI-SearchBot, GPTBot, ChatGPT-User)](https://platform.openai.com/docs/bots) - [OpenAI Help Center: Searching the web with ChatGPT](https://help.openai.com/en/articles/9237897-chatgpt-search) - [Google Search Central: Google Search's guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: Competitor analysis for AI search: how to see who AI names instead of you, and which tools help Canonical: https://whatisgeo.app/competitor-analysis-for-ai-search/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-27 > Ask real buyer questions across several assistants, record who is named, cited and how they are described, then compare with your own pages. Tools repeat this at scale; no 'best rated' tool can be verified. The useful output is a spreadsheet you built yourself: who each assistant names for your buyers' questions, which pages it cites, and how it describes each brand. Do that by hand first. Then you know what a tool would need to repeat, and you can judge whether its numbers hold up. ## What competitor analysis means when the answer is written by AI Classic competitor analysis is about rankings: who sits above you for a keyword, and why. The pages that currently rank for this topic mostly follow that habit. They are generic lists of SEO competitor tools, ordered without a stated method. They are a reasonable starting point for keyword and link data, but they answer a different question. An AI answer has no page one. It is one written response that names you or leaves you out. So you stop measuring position and start measuring presence: - **Share of mentions:** across all the answers you collected, how often each brand is named. - **Cited sources:** which pages the answer links to next to each mention. - **Description:** the words used about each brand, and whether they are accurate. - **Reasons:** what those cited pages have that yours lack. Most tool lists stop before the last of these, and it is the one that tells you what to do next. ## What to record for each competitor For every answer, log these for each brand that appears, and for yourself: 1. Named or not, and roughly where in the answer. 2. The URLs cited next to the mention. 3. The type of each cited source: the brand's own site, a review or comparison site, a forum, a news article. 4. The description: category, strengths, price positioning, caveats. 5. The tone: recommended, listed neutrally, or warned against. A hand count you can check line by line is worth more than a dashboard figure you cannot audit. The source-type column often tells you the most. If assistants keep citing the same few third-party comparison pages, you have learned where the answer really comes from. ## A manual method you can run in an afternoon You need a spreadsheet, a list of questions and free access to a few assistants. 1. **Write the questions.** Our suggestion is 15 to 25 that a real buyer would type: "best X for Y", "X vs Y", "is X worth it", "alternatives to X". Use the buyer's words, not your marketing terms, and mostly leave your brand name out. 2. **Run them.** Use ChatGPT, Perplexity and any other assistant your buyers use, and run the same queries in Google Search to see whether an AI Overview appears. Google says AI Overviews often don't trigger, and that AI Overviews and AI Mode may use different models, so the responses and links they show will vary. 3. **Record.** Paste each answer into the sheet and fill in the columns above. 4. **Repeat.** Run every question again, ideally in a fresh session. Answers can vary between runs and between users, so one pass is an anecdote. Our suggestion, in line with [how to measure AI visibility](/how-to-measure-ai-visibility/), is at least three runs per question to start. 5. **Tally.** Count mentions per brand as a rate across runs, then group the citations by source. If you also want the wider checks (crawler access, how AI describes your own business, source mapping), the [AI visibility audit](/ai-visibility-audit/) walks through them, and this sheet fits straight into it. ## Reading the results: why a competitor gets named Once the sheet is full, look for reasons as well as counts. **Open the cited pages.** Does the competitor's page answer the buyer's exact question directly, with specifics? Compare it with your equivalent page. Often the gap is plain: they have a page for that question and you don't. **Check access.** These are documented platform controls you can read in anyone's public robots.txt file, at the root of their domain: | Platform | Bot that governs search answers | Separate training bot | |---|---|---| | ChatGPT | OAI-SearchBot (OpenAI says sites that opt it out will not be shown in ChatGPT search answers, though can still appear as navigational links) | GPTBot | | Perplexity | PerplexityBot | Perplexity says PerplexityBot is not used for foundation model training | | Claude | Claude-SearchBot (Anthropic says disabling it may reduce visibility in search results) | ClaudeBot | | Google AI Overviews and AI Mode | Googlebot (Google says robots.txt for Googlebot is the control for Search, including its AI features) | Google-Extended covers some of Google's other systems | If a competitor allows these bots and you block them, that is worth fixing first. The check has limits. Blocking can also happen at a CDN or firewall, which robots.txt won't show, and user-triggered fetchers (ChatGPT-User, Perplexity-User) are handled differently: OpenAI says robots.txt rules may not apply to them, and Perplexity says its fetcher generally ignores them. The [AI crawlers and robots.txt](/ai-crawlers-robots-txt/) guide covers the detail. **Look at third parties.** If assistants cite review or comparison sites that mention the competitor and leave you out, the fix sits outside your own domain. Know that before you rewrite a dozen pages. **Be careful about causes.** A competitor named often may simply be older, better known or a clearer fit for the category. The sheet lets you form hypotheses. It cannot prove them. ## What tools add, and what they cannot tell you Tools add scale and repetition. They run many prompts on a schedule across several assistants and keep the history. That saves real time once your manual run has shown which questions matter. They cannot explain a result. A tool can report that a competitor appeared in a share of its sampled answers. It cannot show why the model chose them, and its prompts may not match your buyers. Any competitor share-of-voice figure is the vendor's own estimate, shaped by the vendor's choice of prompts, models and sampling. Quote it as theirs, and if you quote a vendor's figure or price anywhere, take it from the vendor's own page and date the check. Google's documentation sets one more limit. It says the same foundational SEO practices apply to AI Overviews and AI Mode, with no additional technical requirements and no specific optimisation required. Search Console counts traffic from these features inside the Performance report's Web search type, as part of your overall search traffic. It shows your own site only, so it will not show you rivals. ## Kinds of tool compared The table groups tools by type and does not rank them, because we have no rating method we could defend. For named options and how they differ, see [generative engine optimisation tools](/generative-engine-optimization-tools/). | Tool type | What it does | Question to ask the vendor | |---|---|---| | Traditional SEO suites | Keyword, ranking and backlink data on competitors | Does any of this reflect what AI assistants actually say? | | AI answer trackers | Run prompts across assistants, log brands and citations | Which prompts, which models, how many runs, and who chose them? | | Brand mention monitors | Track mentions across the web and social media | Do they link mentions to AI answers, or only the open web? | | Log and crawler analysers | Show which bots visit your own site | Do they identify AI bots by their published names? | | Free platform reports | Show your own performance on one platform | Do they show competitors at all, or only you? | Read the table as a map of what each type can and cannot see. ## Questions to ask before trusting any tool's competitor numbers 1. How are prompts chosen, and can I supply my own? 2. Which assistants and model versions are covered, and how often are they run? 3. How many runs per prompt, given that answers vary? 4. How is the share-of-voice figure calculated? 5. Can I export the raw answers and citations to check them? 6. What happens to the history when a model changes? If a vendor can't answer plainly, treat its numbers with caution. The guide to [AI search optimisation tool data accuracy](/ai-search-optimization-tool-data-accuracy/) explains why figures differ between tools. ## What to do with the findings Sort the sheet into three lists: 1. Questions where a competitor is named and you are absent. 2. Questions where you are named but described wrongly. 3. Questions where nobody is named well. These are your openings. Fix access problems first. Then build or improve the pages that should answer the missing questions, and correct inaccurate descriptions wherever they come from. Afterwards, rerun the same questions under the same conditions so the comparison is fair. The [measurement section of the main guide](/guide/#measurement) explains how to read changes over time. Keep expectations modest. The GEO paper (Aggarwal et al., KDD 2024) tested content changes on its own benchmark of queries and sources and reported visibility gains under those test conditions. It also found that how well each change worked varied across domains. Treat it as a source of ideas to test on your own pages. It does not predict what will happen on your site, and nobody can guarantee that an assistant will name you. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once your manual sheet shows which buyer questions matter, it reruns them across ChatGPT, Claude, Gemini and Perplexity and records whether your company is named and cited, so you don't have to repeat the copy and paste by hand. ## Questions people ask ### How do I find out which competitors ChatGPT and Perplexity recommend instead of me? Ask them the questions your buyers ask, without naming your business, and record who is named and which pages are cited. Run each question more than once, because answers vary between runs and users, and count how often each brand appears. ### Do traditional SEO competitor tools show AI search competitors? Not on their own. Keyword and backlink data describes search results and links, and an AI answer is a separate output you have to check separately. That data can still help explain why a competitor's cited pages are strong. ### Is there a best rated tool for AI search competitor analysis? We know of no rating method we can verify, so we make no 'best rated' claim. Choose by what you need to see, and by how plainly a vendor answers questions about its prompts, models, number of runs and raw data. ### How often should I repeat a competitor check in AI answers? No platform documents a schedule. Our suggestion is to rerun the same question set at regular intervals and after you change pages, keeping the conditions the same, so you compare like with like and can see a trend through the noise. ### Can I copy what a competitor does to get cited? You can learn from it. Look at what their cited page does for the buyer's question, then make a better page for the same question. Copying tells you nothing about which element mattered, and nothing guarantees a citation. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic (Claude Help Center): Does Anthropic crawl data from the web?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024 (arXiv)](https://arxiv.org/abs/2311.09735) --- # Guide: Do Google AI Overviews reduce clicks to websites? What the evidence shows Canonical: https://whatisgeo.app/do-google-ai-overviews-reduce-clicks/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > On average, yes: every study we checked finds fewer clicks when an AI Overview appears, but estimates vary widely with method and query mix. Google disputes the framing, and no study predicts your site's traffic. The studies that measure this agree on the direction and disagree on the size, because each one measures something different. The question that matters for you is whether it is happening to your queries, and only your own Search Console data can start to answer that. ## The direct evidence: what the main studies measured The figures you will see quoted (34.5%, 38%, 42%, 58%) come from studies with very different designs. Here they are side by side, sorted by who produced them. | Study | Type | Method | Market | Data period | What it measured | Main limits | | --- | --- | --- | --- | --- | --- | --- | | Pew Research Center, published 22 July 2025 | Independent research | Browsing data from 900 adults on a consenting panel: 68,879 unique Google searches, 12,593 of them with an AI summary | US | March 2025 | Visits where the user clicked a traditional result: 8% with an AI summary, 15% without | Observational, so it compares different searches; one month; results pages were re-collected in April 2025 | | Agarwal (Indian School of Business) and Sen (Carnegie Mellon), SSRN working paper posted 3 April 2026 | Independent research | Randomised field experiment: a Chrome extension hid AI Overviews for one group of users | US | January to February 2026 | Outbound organic clicks on searches where an AI Overview appeared: 39.8% lower in the current abstract | Working paper, not peer reviewed; desktop Chrome only; paid, self-selected participants | | Ahrefs, 17 April 2025 | Vendor data | 300,000 keywords from its own database; average desktop click-through rate from aggregated Search Console data | Not stated in the post | March 2024 vs March 2025 | Position one click-through rate on AI Overview keywords against a forecast: about 34.5% lower | Correlational; desktop only; informational keywords only | | Ahrefs update, February 2026 | Vendor data | Same method | Not stated in the post | December 2023 vs December 2025 | Position one click-through rate: about 58% lower, with smaller drops further down (19.4% at position 10) | As above, over a two-year gap that includes many other Google changes | | Define Media Group, reported by Search Engine Land on 12 March 2026 | Industry data from a publisher consultancy | Search Console clicks across its client portfolio (64 sites, according to Search Engine Land) | Publisher clients, including major US news sites | Q1 2023 to Q4 2025 | Total organic search clicks: down 42% in Q4 2025 against the pre-AI Overviews average | Before and after on one portfolio; cannot separate AI Overviews from core updates and other changes; publishers only | Some details matter more than the headlines. **Pew** also found that people clicked a link inside the AI summary itself on just 1% of visits to pages with one. They ended their browsing session on 26% of pages with a summary, against 16% without. Summaries appeared on 18% of the searches in the dataset, and on 60% of searches that began with words such as who, what, when or why. **The Agarwal and Sen experiment** is the only randomised test we found, which is why it is the strongest evidence of cause. Search Engine Journal's report on the first version (27 April 2026) described 1,065 participants recruited through Prolific, AI Overviews appearing on 42% of their queries, and outbound clicks rising from 0.38 to 0.61 per search when AI Overviews were removed. It gave the reduction as 38%. The abstract, last revised 8 July 2026, now gives 39.8%. According to the same report, the effect came from AI Overviews shown at the top of the page, and sponsored clicks and overall search frequency did not change. **Ahrefs** is a vendor that sells SEO tools, so its figures are its own. Its method compares how click-through rates changed for keywords that gained an AI Overview with how they changed for similar informational keywords that did not, and treats the difference as the effect. ## Why the headline percentages disagree They are percentages of different things. Pew counts visits with a click on a traditional result. The experiment counts outbound clicks per search. Ahrefs counts the click-through rate for the page ranked first. Define Media Group counts all organic clicks to its clients' sites. The designs differ too. Only the experiment randomly assigns who sees an AI Overview. Pew compares searches that had a summary with searches that did not, and those searches differ in other ways (longer, question-style searches were far more likely to get one). Ahrefs and Define compare before and after, so anything else that changed in Google over those years is mixed in. Query mix, position and population all move the number. Ahrefs looked only at informational keywords on desktop, and its own figures show the effect shrinking further down the page. The experiment found the effect came from AI Overviews at the top of the page. The samples were a US browsing panel, US desktop Chrome users, a vendor's keyword database and a portfolio of news publishers. Time matters as well. Ahrefs' own estimate moved from 34.5% to 58% between its two runs. Note too that the two 42% figures are unrelated: one is how often AI Overviews appeared in the experiment, the other is Define Media Group's traffic decline. Do not average these numbers. Ask what each one was a percentage of. ## What Google says about clicks from AI Overviews These are Google's own claims about its own product, and they are labelled as such here. - In its Search Central guide to AI features, Google says that when people click from results pages with AI Overviews, "these clicks are higher quality (meaning, users are more likely to spend more time on the site)." - In August 2025, Google's head of Search, Liz Reid, wrote that total organic click volume from Google Search to websites had been "relatively stable year-over-year" and that Google was sending slightly more quality clicks, defined as clicks where users do not quickly click back. The same post called third-party reports of dramatic declines inaccurate and "often based on flawed methodologies". It also said traffic was shifting, with some sites losing and others gaining. - Google's Nick Fox later said Google sends "billions of clicks to websites every week through AI features in Search alone", as reported by Search Engine Land, which noted that Google has not shared the click data behind it. Google's claims and the studies can both be partly true. Fewer clicks per search with an AI Overview could still add up to stable totals if people search more and clicks move to different sites. The quality claim has had one independent test so far: the Agarwal and Sen paper found that engagement on websites, once someone clicked, did not differ between the groups. ## Why a click drop is not the same as a visibility drop Your page can be shown in an AI Overview and still get fewer clicks than it used to. Google's documentation sets out how Search Console handles this: - Clicks and impressions from AI Overviews and AI Mode are included in the Performance report under the Web search type, with no click figure for AI features on its own. - Clicking a link to an external page in an AI Overview counts as a click. An AI Overview takes a single position, and every link in it is given that same position. - Since 31 August 2026, every site has had a generative AI performance report. It shows impressions in AI Overviews and AI Mode by page, country, device and date. It does not show clicks. So you can now see which pages Google shows in AI features, but not how many clicks those features send. A falling click line could mean lower rankings, seasonality, a change in what people search for, or an AI Overview answering the question. For the full tracking setup, see [how to track AI Overviews](/how-to-track-ai-overviews/). ## How to check whether AI Overviews affect your own site This is our recommended approach, not a Google procedure. 1. **Compare like with like.** Set the Performance report to compare a recent period with the same period last year, so seasonality does not mislead you. 2. **Find the exposed pages.** In the generative AI performance report, note which pages get AI feature impressions. In the main Performance report, compare those pages' clicks, impressions, click-through rate and position with your other pages. 3. **Split by query type.** Separate informational and question-style queries from transactional and navigational ones. Pew's data suggests question-style searches are far more likely to get an AI summary. 4. **Separate branded from non-branded.** Filter queries containing your brand name. Searches for your own name are a useful comparison group. 5. **Check position before blaming AI.** If your average position fell, a ranking change may explain the drop on its own. 6. **Note other changes.** Write down Google updates and your own site changes (migrations, redesigns, removed pages) in the same period. If impressions and position hold steady while clicks fall on informational, non-branded queries for pages that get AI impressions, that fits an AI Overview effect. It does not prove one. Our [measuring progress](/guide/#measurement) section covers how this fits with tracking AI answers more widely. ## What to do if your clicks have fallen Work out the cause first. If rankings fell, that is an SEO problem to fix before anything else. If the pattern points to AI Overviews, these are sensible steps (our recommendations): - **Look at value, not only clicks.** Google's own guide suggests tracking conversions and time on site in a tool such as Google Analytics. A smaller number of visits that convert may matter more than the raw count. - **Make sure you can be cited.** Google says a page only needs to be indexed and eligible to show with a snippet to appear as a link in AI Overviews, with no extra technical requirements. [How to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/) covers the rest. - **Think before opting out.** Snippet controls such as nosnippet limit what Google can show, and they also limit how your page appears in search results generally. - **Look at which content still earns clicks.** Define Media Group found breaking news held up for its publishers while evergreen content fell. That finding applies to news sites, but the habit of checking which content types are holding up applies to any site. Nobody can guarantee a click or a citation in an AI Overview, and any promise of a recovery should be treated with suspicion. ## What nobody can tell you yet No study gives a number that applies to a single site. Most of the evidence is from the US, much of it is desktop only, and the strongest study is still a working paper. Google has not published click data for its AI features, and Search Console does not show AI Overview clicks separately. AI Overviews also keep changing, as the move between Ahrefs' two figures suggests. More background on the feature itself is in [what are Google AI Overviews](/what-are-google-ai-overviews/), and the wider set of figures is in our [GEO statistics guide](/generative-engine-optimization-statistics/). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Search Console only covers Google, so if you also want to know whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions, it tracks that. ## Questions people ask ### Do Google AI Overviews reduce clicks to websites? Every study we checked finds fewer clicks on searches where an AI Overview appears. The only randomised experiment we found measured 39.8% fewer outbound organic clicks on those searches. Google says overall click volume has been relatively stable, and none of these studies predicts what happens to a particular site. ### Why do studies give different percentages for the click drop? They measure different things. Pew counts visits with any click, the experiment counts clicks per search, Ahrefs counts position one click-through rate and Define Media Group counts total clicks across its clients' sites. The designs, query mixes, devices and time periods also differ. ### Is AI killing website traffic? The evidence shows fewer clicks on searches with an AI Overview. It does not show that every site loses traffic. Google says traffic is shifting between sites, so whether your own traffic is falling is something only your own data can show. ### How can I see whether AI Overviews are hurting my own traffic? Search Console's generative AI performance report shows which of your pages get impressions in AI Overviews and AI Mode, but it does not show clicks. Compare those pages' clicks, click-through rate and position against last year in the main Performance report. Split out branded and question-style queries. ### Are clicks from an AI Overview higher quality? Google says they are, meaning people spend more time on the site, but it has not published the data behind that. The Agarwal and Sen field experiment found no difference in engagement on websites once a click happened. ## Sources - [Pew Research Center: Google users are less likely to click on links when an AI summary appears in the results (22 July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Agarwal and Sen: The Impact of Google AI Overviews on Publisher Traffic and User Experience (SSRN working paper)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6513059) - [Search Engine Journal: Study Confirms Google AI Overviews Cut Organic Clicks 38%](https://www.searchenginejournal.com/ai-overviews-cut-organic-clicks-38-field-study-finds/573145/) - [Ahrefs: AI Overviews Reduce Clicks by 34.5% (vendor study, April 2025)](https://ahrefs.com/blog/ai-overviews-reduce-clicks/) - [Ahrefs: 2026 update to its AI Overviews click study (vendor study)](https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/) - [Define Media Group: BREAKING! News Thrives in the Age of AI](https://www.definemg.com/breaking-news-thrives-in-the-age-of-ai/) - [Search Engine Land: Google AI Overviews cut search clicks 42%: Report](https://searchengineland.com/google-ai-overviews-cut-search-clicks-report-471497) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google: AI in Search is driving more queries and higher quality clicks (Liz Reid, 6 August 2025)](https://blog.google/products/search/ai-search-driving-more-queries-higher-quality-clicks/) - [Search Engine Land: Google says AI Search features send billions of clicks to websites each week](https://searchengineland.com/google-says-ai-search-features-sending-billions-of-clicks-to-websites-each-week-482599) - [Search Console Help: What are impressions, position, and clicks?](https://support.google.com/webmasters/answer/7042828) - [Search Console Help: Generative AI performance report (Search)](https://support.google.com/webmasters/answer/16984139) - [Google Search Central Blog: Introducing Search Generative AI performance reports in Search Console](https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports) --- # Guide: Does Reddit help AI visibility? What is documented, what studies show, and what to do Canonical: https://whatisgeo.app/does-reddit-help-ai-visibility/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Reddit threads do appear in AI answers, and Google and OpenAI both have Reddit data deals. Nobody can promise posting there gets you cited, and vendor citation figures describe their sample, not yours. What has actually been documented is a short list: two licensing deals, Google's general guidance on its AI features, and Reddit's own rules. Most of the rest is vendor measurement or advice. This page labels every claim as one of three kinds. **Documented** means Reddit, Google or OpenAI say it themselves. **Measured** means a study found it, under conditions we state. **Recommended** is our own advice. ## Why people think Reddit matters for AI answers The belief has three roots. Google and OpenAI announced data deals with Reddit in 2024. Google said it wanted Reddit discussions to be easier to find across its products. Vendors have also published tracking data showing Reddit threads among the sources AI tools cite. Taken together, they suggest a simple story: Reddit feeds the AI tools, so Reddit is where you should be. The instinct is reasonable. People ask AI tools questions shaped like forum posts: which tool is better, has anyone tried this, is it worth the money. Forums are full of answers in that shape, which is a good reason to pay attention to Reddit. It is a weak reason to rebuild your marketing around one site, because none of those three roots tells you anything about your brand, your market or your customers' questions. ## What is documented: Reddit's deals and how platforms use forum content ### The Google deal (February 2024) On 22 February 2024 Google announced an expanded partnership with Reddit. Google's post says it "now has access to Reddit's Data API", which it describes as giving it fresher, structured information and better signals for understanding Reddit content. It says the deal "will facilitate more content-forward displays of Reddit information" across Google products, and that it "does not change Google's use of publicly available, crawlable content" for indexing, training or display. Reddit also got a Google Cloud partnership to build AI features with Vertex AI. ### The OpenAI deal (May 2024) On 16 May 2024 OpenAI and Reddit announced a partnership. The joint post says OpenAI "will access Reddit's Data API" and "bring enhanced Reddit content to ChatGPT and new products", so that its tools can better understand and show Reddit content, "especially on recent topics". Reddit gets to build AI features on OpenAI's models, and OpenAI became a Reddit advertising partner. OpenAI's post discloses that Sam Altman is a Reddit shareholder. Neither announcement says how much weight Reddit content gets in an answer, or that any particular thread will be quoted. A licence lets a company use content. It does not decide which content ends up in an answer. ### What Google documents for its AI features Google's page on AI features and your website says there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary" beyond normal Search eligibility. It describes "query fan-out", where the features run several related searches across subtopics and data sources. That is one reason they can show a wider set of links than a classic results page. The page does not single out forums or Reddit. Google also publishes discussion forum structured data (`DiscussionForumPosting`) for forum-style sites where people "collectively share first-hand perspectives". It says this helps Search "better identify online discussions across the web". The markup is for sites that host user posts. Google says not to use it for content written by a site's own publishers, so a business cannot add it to its own articles. ### Reddit's own rules For a business, Reddit's rules matter more than any of the above, because breaking them gets posts removed and accounts banned. From Reddit's own pages: - Rule 2 asks users to "participate authentically in communities where you have a personal interest" and not to spam or manipulate content. Every community also has its own rules, which its moderators enforce. - Rule 5 says you do not have to use your real name, but you must not "intentionally mislead others" or impersonate anyone. - The spam policy says that if your contributions are "primarily of links to a business that you run, own, or otherwise benefit from", you should be thoughtful about how often you post, or consider Reddit's self-serve advertising. It adds that moderators decide what counts as spam in their community. - The disrupting communities policy bans vote manipulation, including using multiple accounts or coordinated groups to move votes up or down. - The manipulated content policy gives "an organization using multiple, seemingly unconnected accounts" to influence an outcome as an example of a violation. It also asks for AI-generated content to be labelled. Reddit's public content policy says Reddit licenses public posts and comments, and that its main data licensees include "large language model makers" and companies that help brands monitor trends. It says Reddit does not license private messages, deleted posts or content from private or quarantined communities. ## What the studies say, and what they cannot show ### Vendor tracking data SE Ranking, an SEO software company that sells AI visibility tracking, published a Reddit study written by a member of its staff. By its own account, it tracked 482 Reddit posts across 135 subreddits for about three months. Across six checks it recorded 7,307 citations in Google AI Overviews, AI Mode, ChatGPT and Gemini. It reports that the 42 posts that appeared in all six checks (9% of those tracked) produced roughly 48% of the citations. These are SE Ranking's own figures. The post does not say how many prompts were tracked or how they were chosen. Its author calls the findings correlations and writes that "none of this is a guarantee". The pattern is useful to know: on a given topic, a small set of long-lived threads may be cited again and again. It does not show that a new thread you post will join that set. When you see any vendor figure about Reddit, check four things: 1. **The sample.** Which prompts, which topics, and how many? Results from consumer topics say little about a narrow B2B niche. 2. **What was counted.** A citation shows that a source was listed. It does not show how much that source shaped the answer. 3. **When and where.** Answers vary between tools, between runs and over time. 4. **Cause.** Reddit may appear often because of the questions that were asked. That does not make it a lever you can pull. Our [GEO statistics guide](/generative-engine-optimization-statistics/) applies the same checks to other numbers that circulate. ### Academic research The best-known paper is Aggarwal et al., *GEO: Generative Engine Optimization*, published at KDD 2024. It tested nine ways of rewriting a page's own content, such as adding quotations, statistics or cited sources. The tests used a benchmark of 10,000 queries on the authors' own GPT-3.5 based engine, plus a 200-query test on Perplexity. Under those conditions, it reports visibility gains of up to 40% for some methods. It tested changes to pages that were already among the sources. It did not measure Reddit activity, brand mentions on forums, or the effect of posting somewhere new. ## Reddit versus your own site and other sources AI cites Reddit is one source among many, and which sources an answer uses depends on the question. Google's own documentation describes its AI features searching across subtopics and data sources to build a response. For a business, the differences look like this: | | A Reddit thread | Your own site | |---|---|---| | Who controls it | Other users and the community's moderators | You | | Can you correct it | Only by replying, within community rules | Yes, directly | | What it is good for | Real users' experiences and comparisons | Your facts: prices, products, policies | | Main risk | Removal, bans, backlash if promotion is spotted | Pages going out of date | We recommend keeping your own site as the source of truth. If a thread describes your product wrongly, publish a clear, accurate page of your own that anyone, including an AI system, can find. An honest reply on the thread can help too, if the community allows it. For the wider work, see [how to improve AI visibility](/how-to-improve-ai-visibility/), [how to rank in ChatGPT](/how-to-rank-in-chatgpt/) and the [principles](/guide/#principles) in our main guide. ## What to do, and what to avoid Everything in this section is our recommendation. None of it is a documented ranking factor. **Do:** 1. Take part under an identity that is honest about who you are. Reddit does not require your real name, but its rules forbid misleading people. 2. Say so when you work for, own or benefit from a business you mention. 3. Answer questions where you have real expertise, including those where the honest answer has nothing to do with your product. 4. Read each community's rules before posting, and ask the moderators if you are unsure. 5. Watch threads that already mention you or your competitors. They exist whether or not you post, and they show you what buyers actually think. 6. Fix your own pages when a thread shows people misunderstanding you. **Avoid:** 1. Praising your brand from several accounts, or paying people to do it. This is astroturfing, and it falls under Reddit's rules on vote manipulation and misleading behaviour. 2. Posting mostly links to your own business. 3. Posting AI-generated content without saying so. Reddit asks for it to be labelled. 4. Promising yourself or a client that any of this will lead to citations. No one can guarantee inclusion in an AI answer. The risks are lopsided. A useful thread might get cited, or might not. An exposed astroturfing campaign is public, searchable and attached to your name. ## How to check whether Reddit affects your own visibility An industry-wide percentage cannot tell you about your own market. Test it yourself: 1. Write down the questions your customers ask before they buy, in their own words. 2. Ask those questions in the AI tools you care about, and note which sources each answer cites. Repeat on different days, because answers vary. 3. Note how often Reddit threads appear, which ones they are, and whether they mention you, a competitor or nobody. 4. Read those threads. Are they accurate about you? Is there a genuine question you could help with, within the community's rules? 5. Keep dated records, change one thing at a time, and distrust any result you cannot repeat. If Reddit threads keep appearing for your questions, that is worth acting on, carefully and honestly. If they do not, you have saved yourself the effort. Our guide to [measuring AI visibility](/how-to-measure-ai-visibility/) sets out a fuller method, and the [measurement section](/guide/#measurement) of the main guide covers the basics. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you have checked a few buyer questions by hand, it tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company over time. ## Questions people ask ### Does Reddit help AI visibility? Reddit threads can appear as sources in AI answers, and Google and OpenAI both have access to Reddit's data. Nobody has shown that posting there will get your business cited, so test your own customers' questions before you spend time on it. ### Does AI get information from Reddit? Google and OpenAI both announced access to Reddit's Data API in 2024, in February and May. Reddit's public content policy names large language model makers among its main data licensees. Neither says how much weight any thread gets in a particular answer. ### Why does Google show so many Reddit results? Google has not given a single reason. Its February 2024 announcement said the Reddit partnership would allow "more content-forward displays of Reddit information" across its products, and it publishes forum structured data to help Search identify discussions across the web. ### Can I get my brand cited by posting on Reddit? Nobody can guarantee it. Vendor data such as SE Ranking's suggests a small set of long-lived threads collects most of the citations on a topic, and a new post may never join that set. Honest, useful participation is still worth doing for its own sake. ### Is it safe to promote my business on Reddit? Only within Reddit's rules. They ask for authentic participation, and the spam policy asks people whose posts are mostly links to their own business to post less often or use Reddit's advertising. Using several accounts or manipulating votes breaks the rules. Check each community's rules first. ## Sources - [Google: An expanded partnership with Reddit (22 February 2024)](https://blog.google/inside-google/company-announcements/expanded-reddit-partnership/) - [OpenAI and Reddit: OpenAI and Reddit Partnership (16 May 2024)](https://openai.com/index/openai-and-reddit-partnership/) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central: Discussion forum structured data](https://developers.google.com/search/docs/appearance/structured-data/discussion-forum) - [Reddit Rules](https://redditinc.com/policies/reddit-rules) - [Reddit Help: Spam](https://support.reddithelp.com/hc/en-us/articles/360043504051-Spam) - [Reddit Help: Disrupting Communities](https://support.reddithelp.com/hc/en-us/articles/360043066412-Disrupting-Communities) - [Reddit Help: Manipulated Content and Misleading Behavior](https://support.reddithelp.com/hc/en-us/articles/41180423371156-Manipulated-Content-and-Misleading-Behavior) - [Reddit Help: Public Content Policy](https://support.reddithelp.com/hc/en-us/articles/26410290525844-Public-Content-Policy) - [SE Ranking (vendor): Reddit and AI Search, What Patterns Show](https://seranking.com/blog/reddit-ai-search/) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: Does schema markup help with AI search? What is documented, what is tested, and what to do Canonical: https://whatisgeo.app/schema-markup-for-ai/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Google says AI Overviews and AI Mode have no additional technical requirements, so schema is not required. It does help machines read a page, so add accurate markup for the facts you already show. Schema markup labels the facts on a page so software can read them without guessing. It cannot win you a place in an AI answer by itself, and nobody can promise you that place. Most claims you will read about it fall into one of three groups: what the platforms document, what research has tested, and what is simply sensible practice. This page keeps the three apart and tells you which is which. ## What schema can and cannot do for AI answers Schema can describe your business, your products and your articles in a fixed vocabulary that search engines understand. Google says it uses that markup to understand pages. That is a real benefit, and it costs little to get right. No platform we cite here says that schema decides which sources an AI answer uses. Google's documentation for AI Overviews and AI Mode lists no markup requirement at all. So treat schema as a way to make your facts easy to read. It does not replace being a page worth citing. If you want the bigger picture of how AI answers choose sources first, start with [how AI answers happen](/guide/#how-it-works). ## What the platforms document ### Google on AI Overviews and AI Mode Google's guide to AI features says that to be eligible as a supporting link in AI Overviews or AI Mode, "a page must be indexed and eligible to be shown in Google Search with a snippet, fulfilling the Search technical requirements. There are no additional technical requirements." It adds: "While specific optimization isn't required for AI Overviews and AI Mode, all existing SEO fundamentals continue to be worthwhile." Structured data appears once in that list of fundamentals: "Making sure your structured data matches the visible text on the page." In other words, Google's AI guidance mentions schema as something to keep accurate. It never presents it as a way in. Our guide to [ranking in Google AI Overviews](/how-to-rank-in-google-ai-overviews/) covers the rest of that list. ### Google on structured data in general Google's introduction to structured data says: "Google uses structured data that it finds on the web to understand the content of the page, as well as to gather information about the web and the world in general." On format, it says "we recommend using a format that's easiest for you to implement and maintain (in most cases, that's JSON-LD)", and adds that JSON-LD, Microdata and RDFa are "equally fine for Google" if the markup is valid. Google's general structured data guidelines set the conditions for rich results (the enhanced listings such as review stars). The guidelines "must be followed in order to be eligible for appearance as a rich result", and breaking a quality guideline "can prevent syntactically correct structured data from being displayed as a rich result". Following the rules makes you eligible. Being eligible does not mean the rich result will appear. ### FAQ and HowTo rich results Many schema guides still promise FAQ rich results. Google's own changelog says the FAQ rich result "will no longer appear in Google Search results starting May 7, 2026", and Google removed its FAQ documentation on 15 June 2026. The HowTo rich result went in September 2023, when Google said it "is no longer shown in search results". So any advice that depends on these features is out of date. ### Schema.org Schema.org maintains the vocabulary itself. Its getting-started guide describes "a collection of shared vocabularies webmasters can use to mark up their pages in ways that can be understood by the major search engines". It defines the types, such as Organization, LocalBusiness and Product, along with their properties. It says nothing about AI citations. ## What the published research covers, and what it does not People often cite Aggarwal et al., "GEO: Generative Engine Optimization" (KDD 2024) as evidence for all kinds of GEO tactics. The paper tested changes to the text of a page. Its strongest methods were Cite Sources, Quotation Addition and Statistics Addition, which the authors describe as adding citations from reliable sources, credible quotes and relevant statistics to the content. They measured the results on their own benchmark of queries (GEO-bench). Schema markup was not one of the methods tested. The paper tells you nothing about schema either way. You will also find vendor studies quoting figures for how much more often marked-up pages get cited. We do not repeat those figures here. If you meet one, remember that it belongs to the vendor and comes from a method you usually cannot check. Before trusting any schema test, ask three things: 1. How many pages and queries were tested? 2. Did only the markup change, or did the content, links and ranking position change too? 3. Which platform was tested, and when? A result from one engine at one moment may not hold after the next update. A test on a handful of pages, or one that changed several things at once, cannot show that schema caused the result. ## Which schema types are worth adding, and which to skip This section is our recommendation, not a documented platform rule. For most small businesses we would add these, and only where the page already shows the information: | Type | Where it goes | What it labels | |---|---|---| | Organization | Home or about page | Your name, logo, website and contact details | | LocalBusiness | Home or contact page, if customers visit or you serve an area | Address, phone, opening hours | | Article | Guides and blog posts | Headline, author, publication and update dates | | Product or Service | Pages that sell something | What is offered and, if shown, the price | LocalBusiness is a more specific kind of Organization in the Schema.org vocabulary, so a local firm can use it in place of plain Organization. Google also suggests using "the most specific applicable type" Schema.org offers. FAQPage is still a valid Schema.org type. Add it only for real questions and answers shown on the page, and do not expect a Google rich result from it. Skip anything that describes what a visitor cannot see: reviews you have not collected, awards you have not won, opening hours you do not keep. More types do not make better markup. Accurate types do. ## How to add it: JSON-LD, one example and a check JSON-LD sits in a single script block, separate from your visible text. Here is a minimal LocalBusiness example. Every value is a placeholder for your own details: ```html ``` Then check it with two free tools: - **Google's Rich Results Test** (search.google.com/test/rich-results) shows whether Google can read the markup and whether the page qualifies for any rich result. - **The Schema Markup Validator** (validator.schema.org), run by Schema.org, checks the markup against the vocabulary, whether or not Google uses that type. Where you put the markup also matters. Google says it can read JSON-LD that is injected into the page by JavaScript. Other crawlers may not handle it the same way, and we have not seen documentation from every AI company on this point. As a precaution, we put the markup in the HTML the server sends, so it does not depend on scripts running. This is our own practice, not a documented requirement. Whether a crawler can reach the page at all is a separate question, covered in [crawler access](/guide/#crawler-access). ## Keep the markup honest: match the visible page This is the point with the most documentation behind it. Google's guidelines say: "Don't mark up content that is not visible to readers of the page." Its AI features guide repeats the idea: make sure structured data "matches the visible text on the page". Fake reviews and misleading markup can lead to a manual action. Google says this takes away rich result eligibility but "doesn't affect how the page ranks in Google web search". In practice: - If the page says you open at nine, the markup says nine. - If a price changes on the page, change it in the markup at the same time. - If a product is out of stock, do not label it available. Markup that contradicts the page gives a machine two versions of the facts and no way to choose between them. When someone edits a page, have them check its markup too. ## Where schema fits in the wider GEO picture Schema labels facts you already publish. It cannot make a thin page useful or make an unknown business well known. The work that usually matters more is clear, specific content, pages that crawlers can reach, and a way to see whether anything changes. Our [principles](/guide/#principles) cover the content side. [Measuring progress](/guide/#measurement) covers how to check results without assuming them. If you want to know how often AI tools mention you at all, start with [what AI visibility means](/what-is-ai-visibility/). Treat schema as housekeeping. It is worth doing and cheap to do well, but it will not stand in for having something worth citing. ## From the same studio These are made by the studio that publishes this guide. - [Hilyt](https://hilyt.it/): If your next question is how AI systems describe your business, Hilyt publishes an open, AI-readable profile alongside the markup on your own site. ## Questions people ask ### Does schema markup help you appear in Google AI Overviews? Google says a page only needs to be indexed and eligible to show with a snippet, and that "there are no additional technical requirements". It does say Google uses structured data to understand pages, so accurate markup can help in that way. Nobody can guarantee a place in an AI Overview. ### Do ChatGPT, Perplexity or Claude read schema markup? This page makes no claim either way, because we have not read documentation from OpenAI, Perplexity or Anthropic that says how their systems use schema markup. If it matters to you, check each provider's own documentation and treat second-hand claims with care. ### Which schema types should a small business add first? We recommend Organization, LocalBusiness if customers visit or you serve a local area, Article on guides and posts, and Product or Service on sales pages. This is our recommendation and not a platform rule. Only mark up what a visitor can see on the page. ### What format should schema markup be in, and how do I test it? Google recommends JSON-LD in most cases. Check the markup with Google's Rich Results Test (search.google.com/test/rich-results) and the Schema Markup Validator (validator.schema.org). ### Is schema markup the same as structured data? In everyday use the two mean almost the same thing. Structured data is any machine-readable description of a page. Schema.org supplies the shared vocabulary most of it uses, and JSON-LD is the usual format you write it in. ### Is FAQ markup still worth adding? Google's changelog says the FAQ rich result has stopped appearing in Search since 7 May 2026, and the HowTo rich result was removed in 2023. FAQPage is still a valid Schema.org type, so marking up real, visible FAQs does no harm, but do not expect a Google rich result from it. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central: Introduction to structured data markup](https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data) - [Google Search Central: General structured data guidelines](https://developers.google.com/search/docs/appearance/structured-data/sd-policies) - [Google Search Central: Latest documentation updates (FAQ and HowTo rich results)](https://developers.google.com/search/updates) - [Schema.org: Getting started](https://schema.org/docs/gs.html) - [Schema.org: LocalBusiness type](https://schema.org/LocalBusiness) - [Aggarwal et al.: GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Google Rich Results Test](https://search.google.com/test/rich-results) - [Schema Markup Validator](https://validator.schema.org/) --- # Guide: Free AI tools for SEO: what is genuinely free, what each one is for, and what to skip Canonical: https://whatisgeo.app/free-ai-tools-for-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-26 > The most useful free tools for AI-era SEO are the platforms' own: Google Search Console and Bing Webmaster Tools, plus a manual check of AI answers. A free AI assistant helps with drafting, but check its work. Start with the free reports the search engines provide for your own site. Then add free checking tools and a manual look at AI answers. Almost anything else sold as a "free AI SEO tool" is a capped tier or a trial. Either is fine to use, as long as you know which one you are looking at. ## What "free" means for SEO tools, and why most free tiers are samples "Free" covers three different things, and most listicles blur them: 1. **Free because the platform wants a healthier web.** Search engines run reports and testing tools for site owners without charging for them. 2. **Free tiers.** A cut-down version of a paid product. You get enough to see what it does, but rarely enough to run your whole site through it. 3. **Trials.** The full product with an end date: a sales process with a countdown. None of these is wrong to use. Readers get misled when all three are called "free" without saying which is which. Limits and plan details change, often without notice. We quote a limit only where we read it on the vendor's own page, and we give the date. For everything else, check the vendor page on the day you read this. We also name no "best" tool. Our method for this page is three questions. Does the platform itself say it is free? Does it show your data or someone's opinion? What can it not tell you? ## Free tools that show your real data Set these up first. Everything else depends on them. **Google Search Console.** Google describes it as "a free service offered by Google" that helps you monitor and troubleshoot your site's presence in Google Search. Its Performance report shows clicks, impressions, click-through rate and average position, and you can group the data by the queries people typed. That is your baseline: what people searched for, when Google showed you and whether they clicked. On AI features, Google's documentation says AI Overviews and AI Mode are counted towards the overall data in Search Console. That traffic is already inside your totals. Check the report's filters yourself before assuming you can separate it out. **Bing Webmaster Tools.** Microsoft says you can "get started for free". In February 2026 it announced an AI Performance report in public preview. The report shows how often your content is cited in Microsoft Copilot, in AI-generated summaries in Bing and in some partner integrations. It also shows which URLs were cited and a sample of the phrases the AI used to find them. Microsoft is clear that these counts show citation activity, not ranking or importance. Previews change, so check the report's status when you sign in. Neither tool shows what ChatGPT, Claude, Gemini or Perplexity said about you to a real person. Each one reports only on its own platform. For how to read this data properly, see [how to measure AI visibility](/how-to-measure-ai-visibility/). ## Free tools that check crawling, markup and page problems **Rich Results Test and Schema Markup Validator.** Google's structured data documentation recommends starting with the Rich Results Test to see which Google rich results can be generated for a page. For generic schema.org checks, it points to the Schema Markup Validator, which tests markup "without Google-specific validation". The Rich Results Test also reports whether a robots.txt rule stops Google crawling the page. Both work on a live URL or on a code snippet you paste in. Read the results for what they are. A pass means the markup can be read and is eligible. It does not mean Google or an AI assistant will use it. For the longer version, see [schema markup for AI](/schema-markup-for-ai/). **Your robots.txt, checked against each vendor's own docs.** OpenAI, Perplexity and Anthropic each publish which crawlers they run and how to allow or block them in robots.txt. OpenAI, for example, separates OAI-SearchBot (for ChatGPT search results) from GPTBot (for training its models), so you can allow one and block the other. The vendors define their own bot names, so read their pages, not a list copied from somewhere else. We keep the detail in one place: [AI crawlers and robots.txt](/ai-crawlers-robots-txt/). **A desktop crawler.** Screaming Frog's own product page says the free version of its SEO Spider crawls up to 500 URLs (checked 26 October 2026). Above that you need a paid licence. ## Free ways to check what AI assistants say about you This is a manual job, and at small scale that works fine. Our recommended routine: 1. Write down the questions a customer would really ask, in their own words. 2. Ask them in the assistants your audience uses. 3. Record the answer, the date, whether you were named and which sources were cited. 4. Run the same list again later, without changing it. Why repeat it? AI answers can differ between runs of the same question, so one check is one observation. That is a working rule from practice, not a statistic. A pattern across several runs tells you more than any single answer. When you report a pattern, say how many runs it is based on. Free "AI visibility checker" tools automate a version of this. Before trusting one, ask how many runs sit behind the result, which prompts were used and which model answered. If the tool cannot say, you cannot interpret the score. Our [AI visibility checker](/ai-visibility-checker/) guide covers what those tools can and cannot tell you. ## Using a free AI assistant for SEO tasks, and where it goes wrong A free assistant earns its place in the dull middle of SEO work: drafting title tag options, summarising an exported list of queries, suggesting how to group them and tidying a clumsy paragraph. It goes wrong in predictable places. It will describe your site without being able to see it. It will state a "best practice" with total confidence. It will analyse a pasted spreadsheet and quietly skip the awkward rows. Give it your real data, ask it to show its working and check the output against the source. Google's guidance supports this. Its page on generative AI content says these tools "don't retrieve facts" and instead predict a likely sequence of words. It asks you to fact-check and review AI-generated content before publishing. It also warns that using generative AI to create many pages without adding value for users may breach its spam policy on scaled content abuse. In 2023, Google said it rewards high-quality content "however it is produced". The risk comes from publishing unchecked output at scale. There is more in [is AI content good for SEO](/is-ai-content-good-for-seo/). Free plan limits for ChatGPT, Claude and Gemini change often. Check each vendor's own plans page before you rely on one for regular work. ## Free tiers and trials from paid tools: how to read them Treat a free tier as a demo with a cap. Four questions sort them quickly: 1. Does it show your own data, or a generic sample? 2. Is the cap so low that you cannot finish the job you came for? 3. Does it say how each result was produced? 4. When the trial ends, what can you export and keep? If the answers are "generic", "yes", "no" and "nothing", the tier exists mainly to sell the paid plan. That is allowed. Just do not build a workflow on it. ## A free toolkit by job, in the order we would set it up This is our recommendation. It is ordered by what each step depends on, and it is not a ranking. | Order | Job | Free option | What it cannot tell you | | --- | --- | --- | --- | | 1 | See your real search data | Google Search Console, Bing Webmaster Tools | What other AI assistants say about you | | 2 | Check markup and crawler access | Rich Results Test, Schema Markup Validator, your robots.txt checked against vendor docs | Whether anything will actually be shown or cited | | 3 | Find page problems | The free version of a desktop crawler | Anything past its crawl limit | | 4 | Check AI answers | A fixed list of questions, asked by hand and logged | A stable result from a single run | | 5 | Speed up drafting and analysis | A free AI assistant | Whether its output is true; you check that | Try paid tools last, and only for a gap these steps have shown up. If you skip ahead, you end up tuning a site you cannot measure. ## What free tools cannot do They cannot tell you why something happened. Search Console shows a drop, but working out the cause is still your job. They cannot give you a stable AI visibility score from one check, and free checkers rarely disclose their method. They cannot judge which customers, pages and claims matter to your business. And no tool, free or paid, can guarantee you a place in an AI answer. For the fuller measurement method, see the [measuring progress section](/guide/#measurement) of our main guide. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If checking AI answers by hand stops scaling, it is one option for tracking whether ChatGPT, Claude, Gemini and Perplexity name and cite your company; put the free-tier questions on this page to it as to any other tool. ## Questions people ask ### Are there really free AI tools for SEO? Yes, but read "free" carefully. Google describes Search Console as a free service, Microsoft says you can get started with Bing Webmaster Tools for free, and Google points site owners to the Rich Results Test and Schema Markup Validator for checking markup. Most third-party "free" offers are capped tiers or trials, so check the vendor page on the date you read it. ### Can ChatGPT do SEO for free? A free AI assistant can help with drafting and analysis. It cannot see your Search Console data unless you give it, and Google warns that generative AI predicts likely words instead of retrieving facts, so check everything it produces. Free plan limits change, so check OpenAI's own plans page. ### Is Google Search Console free, and does it show AI Overviews data? Google describes Search Console as a free service. Google's documentation says AI Overviews and AI Mode are counted towards the overall data in Search Console, so that traffic is already in your totals. ### Are free AI visibility checkers accurate? Treat any single result as one observation, because AI answers can differ between runs of the same question. That is a working rule from practice, not a statistic. A checker is only as useful as the method it discloses: how many runs, which prompts and which model. ### Do I need to pay for an AI SEO tool at all? Not to start, in our view. The free first-party reports, markup tests and a hand-run prompt check cover the basics. Pay when you hit a specific gap you cannot close by hand. ## Sources - [Search Console Help: About Search Console](https://support.google.com/webmasters/answer/9128668) - [Search Console Help: Performance report (Search results)](https://support.google.com/webmasters/answer/7576553) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Bing Webmaster Tools: About](https://www.bing.com/webmasters/about) - [Bing Webmaster Blog: Introducing AI Performance in Bing Webmaster Tools Public Preview (February 2026)](https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview) - [Google Search Central: Introduction to structured data (testing tools)](https://developers.google.com/search/docs/appearance/structured-data) - [Search Console Help: Rich Results Test](https://support.google.com/webmasters/answer/7445569) - [Schema.org: Schema Markup Validator](https://validator.schema.org/) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Screaming Frog: SEO Spider (free version and licence, checked 26 October 2026)](https://www.screamingfrog.co.uk/seo-spider/) - [Google Search Central: Guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) --- # Guide: Generative engine optimisation agency: what to ask before you hire one Canonical: https://whatisgeo.app/generative-engine-optimization-agency/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > A GEO agency can do real work on content, crawler access, structured data and measurement, but none can guarantee inclusion in AI answers. Judge one by its method, its measurement and its honesty about limits. Before you talk to any agency, know that the work is mostly ordinary website work you can check, and the platforms decide the outcome. Google says plainly that "indexing and serving isn't guaranteed" even for pages that meet all its requirements, so a good agency sells a method and a way of measuring it, and a bad one sells certainty. This page ranks no agencies. It sets out what the work consists of, which parts rest on platform documentation, which on research and which on practice, and how to check an agency's claims yourself. If you want the background first, start with [what GEO is](/guide/#what-is-geo). ## What a GEO agency can and cannot promise An agency cannot promise that ChatGPT, Perplexity, Claude or Google's AI features will name or cite you. Nobody outside those platforms controls what they generate. That is our position as a site, and it matches what Google documents for its own AI features. What an agency can promise is effort in defined areas, on a schedule you can see: clearer and better sourced content, crawlers that can reach your pages, accurate markup and a measurement plan. That is a fair scope and it can be useful. The test of a pitch is whether it says what will be done, how it will be checked and where it might fail. ## What GEO agency work actually consists of Strip away the branding and the work falls into four jobs. 1. **Content.** Making pages answer real questions plainly, with sources, quotations and figures where they genuinely belong. 2. **Crawler access.** Making sure the bots that feed AI products can read your site, and that you have chosen deliberately which ones to allow. See [crawler access](/guide/#crawler-access). 3. **Structured data.** Marking up pages so machines can read what they are about. This is useful practice, though Google says no special schema.org markup is needed for its AI features. 4. **Measurement.** Working out whether any of it changed anything. See [measuring progress](/guide/#measurement). If a proposal contains a fifth job it cannot explain in a sentence, ask about it. Much of the rest is relabelled SEO. That is no bad thing, because the fundamentals matter, but you should know what you are paying for. The [GEO vs SEO](/guide/#geo-vs-seo) section covers the overlap. ## What is documented, what is researched, what is just practice This is the section sales pages skip. Claims about GEO come from three different places and deserve different levels of trust. ### Documented by the platforms - **Google.** Its "AI features and your website" page says there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary", that SEO best practices remain relevant, and that a page must be indexed and eligible to show with a snippet. - **OpenAI.** OAI-SearchBot is for search and GPTBot is used for training its foundation models. OpenAI says each robots.txt setting is independent, so you can allow one and block the other. ChatGPT-User handles visits a user starts, and OpenAI says robots.txt rules may not apply to those. - **Perplexity.** PerplexityBot surfaces and links sites in Perplexity's search results. Perplexity-User fetches pages when a user asks a question, and Perplexity says it "generally ignores robots.txt rules" because a person triggered the fetch. - **Anthropic.** ClaudeBot collects content that may be used for training, Claude-User fetches pages when people ask Claude questions, and Claude-SearchBot indexes content for search. Each can be controlled in robots.txt, set separately for each subdomain. - **robots.txt itself.** RFC 9309 defines it as rules that crawlers "are requested to honor", and says those rules "are not a form of access authorization". You can check every one of these on the platforms' own pages, linked in the sources below. ### Researched Aggarwal et al., "GEO: Generative Engine Optimization" (KDD 2024), tested content changes such as citing sources, adding quotations and adding statistics, and found those three performed best. The conditions matter. The main tests used the authors' own benchmark of 10,000 queries and a set-up that took the top five Google results for each query and had gpt-3.5-turbo write the answer. They also ran a smaller test on Perplexity.ai with 200 queries. The authors say methods may need to change as engines evolve, and that they did not measure effects on search rankings. Treat it as good evidence that a direction is worth testing on your own pages. It does not show what ChatGPT or Google will do with your site today. ### Practice Everything else: prompt tracking, entity work, which page formats seem to get quoted. Some of it may well be sound. It is still experience and opinion, and an honest agency will label it that way. If an agency cannot tell you which of these three groups a claim belongs to, that tells you something. ## Do you need an agency at all? Possibly not, and sales pages never say so. If you have someone who owns your site's content and someone who can edit robots.txt and markup, you already have most of the skills. Checking robots.txt against the crawler names above, reading Google's AI features page and rewriting key pages so they answer questions clearly is ordinary work for a capable in-house team. An agency can earn its fee when you have a large site and no capacity, no technical help, or you want an outside view on measurement. If your SEO basics are shaky, fix those first: Google says they still apply, and a new layer on a weak foundation is poor value. The [30-day plan](/guide/#getting-started) is a reasonable way to try the basics yourself before deciding. ## Questions to ask before you hire These are our recommendations, drawn from the limits above; no platform requires them. Use what is useful. 1. What exactly will you change on our site, and in what order? 2. Which of your claims come from platform documentation, and which from research or your own experience? 3. How do you decide which AI crawlers to allow or block, and who signs that off? 4. How will you measure results, and what will we record before you start? 5. What would make you tell us the work is not helping? 6. Do you produce pages in bulk, and how do you keep that within Google's spam policies? 7. Who actually does the work, and who do we speak to? 8. What do we keep if we leave? Listen to the shape of the answers. Specific and bounded is good. Vague and confident is a warning. ## Red flags in pitches and proposals This list is also our recommendation, built on the documented limits above. **Guaranteed citations or placements.** Nobody outside the platform can deliver them, and Google says serving is not guaranteed. Treat any guarantee as disqualifying. **Unsourced statistics.** If a figure has no named source you can check, ask where it came from. A number from a vendor's own marketing is that vendor's claim and should be described as such. **"Secret" tactics.** The documented parts of this field are public. A method that cannot be explained cannot be evaluated. **Mass-produced pages.** Google's spam policies define scaled content abuse as generating many pages "for the primary purpose of manipulating search rankings", and say this applies "no matter how it's created", naming generative AI tools as one example. A plan for hundreds of thin pages puts your site at risk. **Selling robots.txt as security.** It is a voluntary standard for crawlers, and some user-triggered fetchers say they may not follow it. Anyone who presents it as a lock has misread it. ## How to judge results: set a baseline first Most buyers hire, wait, and then try to work out whether anything changed. By then there is nothing to compare against. Our recommended method is to record where you stand before any work starts. This is practice; no platform requires it. - The questions that matter to your business, written down word for word. - What each AI tool currently says in answer, saved with the date. - What your analytics and Search Console show. Know what Search Console can and cannot tell you. Google says traffic from AI Overviews and AI Mode is included in the Performance report under the "Web" search type, so there is no separate AI line. Be wary of anyone who presents one as if it came from Google. Then repeat the same checks at agreed intervals: same questions, same tools, same method. Keep your own copy of the records. That is your independent check, and it works whether or not you hire anyone. ## Names you will see: GEO, AEO, LLM SEO, AI SEO Generative engine optimisation, answer engine optimisation, LLM SEO and AI SEO are different labels for work that overlaps heavily; [SEO vs GEO vs AEO](/seo-vs-geo-vs-aeo/) sets them side by side. Do not pay extra for a name. Compare each task list against the four jobs above. If two proposals carry different labels but list the same tasks, they are the same offer. ## How to compare agencies without a "best of" list Many "best agencies" lists are ranked by the publisher's own criteria, sometimes with the publisher near the top. You can build a fairer comparison yourself. 1. Send the same eight questions to each shortlisted agency, in writing. 2. Score each answer on three things: how specific the method is, how clear the measurement plan is, and how openly limits are described. 3. Ask each agency to mark its claims as documented, researched or practice. 4. Open any source they cite and check it says what they say it does. The agency that does best is usually the one whose answers survive being checked. The site's principles for GEO are a handy reference while you do it. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want a repeatable baseline to judge an agency against, it tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions. ## Questions people ask ### What is the best agency for generative engine optimization? No honest answer applies to everyone, and we do not name or rank agencies. The right one depends on your site, your budget and the work you need done, so compare shortlisted agencies on how specific their method is, how clear their measurement plan is and how openly they describe limits. ### Can a GEO agency guarantee my business appears in ChatGPT or Google AI Overviews? No. Nobody outside the platform controls what it generates, and Google says indexing and serving is not guaranteed even for pages that meet all its requirements. Treat any guarantee as a red flag. ### Is a GEO agency different from an SEO agency? Partly. The overlap is large, and Google says SEO best practices remain relevant for its AI features. The extra work is mostly crawler choices across several AI products, content written to be quoted, and new ways of measuring, so compare the scope of work and ignore the label. ### How do I know if a GEO agency is working? Record a baseline before work starts: your key questions, what each AI tool says in answer, and your analytics and Search Console data. Repeat the same checks at agreed intervals and keep your own copy. Google includes AI Overviews and AI Mode traffic in the Performance report's Web search type, so there is no separate AI line to point to. ### Do I need an agency for GEO or can I do it in-house? Often you can do it in-house. Checking crawler access, improving key pages and adding basic markup are within reach of most capable teams; an agency helps most when you lack capacity, technical help or an outside view on measurement. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central: Spam policies for Google web search](https://developers.google.com/search/docs/essentials/spam-policies) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full text (HTML)](https://arxiv.org/html/2311.09735v3) - [OpenAI: Overview of OpenAI crawlers](https://developers.openai.com/api/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) --- # Guide: Generative engine optimisation checklist: what to do, in order, and what each item rests on Canonical: https://whatisgeo.app/generative-engine-optimization-checklist/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > A GEO checklist covers crawler access, indexing, quotable content, identity and measurement. A few steps are platform requirements; the rest is research or sensible practice. None guarantees a citation. Work through it in order: first make sure the crawlers behind AI search can reach your pages, then that those pages are indexed and eligible for a snippet, then make each page easy to quote. Only the first two steps rest mainly on what platforms document; the content step leans partly on one research paper, and most of the rest is our recommendation, labelled as such. ## What a GEO checklist can and cannot do Many GEO checklists are long runs of tick-boxes that never say what each item rests on. You tick everything and still don't know what mattered. A checklist can do three useful things. It can make sure the right systems are allowed to reach your pages. It can make those pages easy to quote accurately. And it can give you a repeatable way to check what happens next. It cannot promise a citation. AI answers are assembled by systems you don't control, and none of the platform documents below offers a guarantee. Google even notes that AI Overviews are only shown when its systems judge them additive to classic Search, "and as such, often don't trigger". If someone sells you guaranteed inclusion, that tells you something about the seller. The same list serves whether you call the work GEO, AEO or LLM SEO. The terms are used loosely and the practical steps overlap almost completely. ## How to read this checklist: documented, researched or recommended Every item carries one of three labels: - **Documented:** a platform states it in its own published guidance. This is the strongest label, though it still describes one platform at one point in time. - **Researched:** a named study found an effect under stated conditions. Useful, but a study is not a live platform. - **Our recommendation:** sensible practice with no source behind it. We think it is worth doing; nobody has proved it works. Guidance gets revised, so read the linked source yourself before acting on our summary of it. ## 1. Access: can the right crawlers reach your pages? **Label: documented.** Start here, because nothing else matters if the door is shut. The main AI companies publish separate crawlers for separate jobs, and you set your preferences for each in robots.txt. | Platform | Search or indexing | Training | User-initiated fetches | |---|---|---|---| | OpenAI | OAI-SearchBot | GPTBot | ChatGPT-User | | Perplexity | PerplexityBot | None listed | Perplexity-User | | Anthropic | Claude-SearchBot | ClaudeBot | Claude-User | | Google | Googlebot | Google-Extended (some other Google systems) | Not covered here | What each platform says, in short: - **OpenAI** says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. Each setting is independent, so you can allow OAI-SearchBot while disallowing GPTBot. Because ChatGPT-User acts on a user's request, OpenAI says robots.txt rules may not apply to it. OpenAI also lists OAI-AdsBot, which only visits pages submitted as ads. - **Perplexity** recommends allowing PerplexityBot if you want to appear in its search results, says PerplexityBot is not used to crawl content for AI foundation models, and says Perplexity-User generally ignores robots.txt because a user requested the fetch. - **Anthropic** says its bots honour robots.txt, and that disabling Claude-SearchBot or Claude-User may reduce your visibility in user search results. - **Google** says robots.txt rules for Googlebot are the control for its AI features in Search, and that crawling should be allowed in robots.txt "and by any CDN or hosting infrastructure". That second part is easy to miss: a firewall or bot-blocking setting at your host can shut out crawlers your robots.txt allows. So check each crawler separately, and check your CDN as well as the file. Search crawlers and training crawlers are different decisions, and a blanket "block all AI bots" rule may remove you from the answers you wanted to appear in. One caveat: robots.txt is defined in RFC 9309, which says crawlers "are requested to honor" its rules and that the rules "are not a form of access authorization". It states your preferences; it does not enforce them. The full walkthrough, with example rules, is in [AI crawlers and robots.txt](/ai-crawlers-robots-txt/). ## 2. Eligibility: are your pages indexed and snippet-eligible? **Label: documented.** Google's guidance on its AI features is direct. To be eligible as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet. Google says there are no additional technical requirements. That makes the first Google check an ordinary one. Is the page indexed? Can it show with a snippet, or has a `nosnippet` or `noindex` rule removed it? Google suggests verifying your site in Search Console to catch technical problems. Fix these before anything cleverer. Bing's webmaster guidelines say Bing and Copilot rely on the same crawling, indexing and ranking foundation as traditional search. They recommend sitemaps with accurate `lastmod` values, and IndexNow to tell Bing when URLs change. They also note that `noarchive` prevents content from being used in Copilot responses, and `nocache` limits Copilot to the URL, title and snippet. Check you have not set either by accident. For Google detail, see [how to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/). ## 3. Content: is each page easy to quote? **Label: partly documented, partly researched, partly our recommendation.** *Documented.* Google says specific optimisation isn't required for its AI features and that existing SEO fundamentals still apply. Its list includes making sure important content is available in textual form, providing a good page experience, making content findable through internal links, and creating helpful, reliable, people-first content. Its helpful-content guidance asks whether content earns trust through "clear sourcing" and evidence of expertise. None of this is specific to AI, and that is the point. *Researched.* Aggarwal et al., in "GEO: Generative Engine Optimization" (KDD 2024), tested nine ways of rewriting source content on GEO-bench, a benchmark of queries the authors built. Their generative engine fed the top five Google results for each query to a language model. Adding citations, quotations and statistics performed best, with reported visibility gains of up to 40% on the authors' own metrics; keyword stuffing did not perform well; and the effect varied by subject area. They also ran the methods on Perplexity.ai. Because it is a benchmark study of rewritten pages, treat it as a reason to test these changes on your own pages. It cannot tell you what any one edit will do on a live platform today. *Our recommendation.* No source claim behind these: - Put the direct answer near the top of the page. - Keep one question per page. - Date your facts and say where each one came from. A page that states its answer plainly, once, with its source named, is easier for a person to quote. We think the same is likely true for a machine, but no platform document we have read says so. More on the content side is in [how to improve AI visibility](/how-to-improve-ai-visibility/). ## 4. Entity: is it clear who you are and what you do? **Label: mostly our recommendation, one documented item.** *Documented.* For Google, keep your Business Profile and, if you sell products, your Merchant Center information up to date. Google lists both in its SEO best practices for AI features. *Our recommendation.* Use the same business name, location and description of what you offer across your site and every profile you control. A system trying to work out who you are has to match scattered mentions to one business, and inconsistency makes that harder. That is our reasoning; no platform documents it as a requirement. It is cheap to do and unlikely to hurt. ## 5. Optional extras: schema and llms.txt, and what is actually documented **Label: documented for what Google requires; otherwise our recommendation.** *Schema.* Google says there are no additional technical requirements for its AI features, so structured data is not a gate. Where you do use it, Google's advice is to make sure it matches the visible text on the page. Use it for its own reasons and keep it honest. *llms.txt.* This is a proposal published at llmstxt.org, with no standards-body status. The proposal notes that thousands of sites publish one and that the AI labs publish llms.txt files for their own developer docs. None of the crawler documents from OpenAI, Perplexity or Anthropic that we read for this page mentions using it, and neither does Google's AI features guidance. It is cheap to add, and that is the whole case for it. The detail is in our [llms.txt guide](/llms-txt/). Both are optional, and neither belongs ahead of access, eligibility or content. ## 6. Measurement: how to check the result **Label: one documented item, the rest our recommendation.** *Documented.* Google says traffic from its AI features is counted in the Search Console Performance report, within the overall "Web" search type. It does not appear there as a separate line. *Our recommendation.* Write down a fixed set of the questions your buyers actually ask. Put the same questions to the same platforms on a regular schedule, and record who is named, who is cited and how your business is described, including when you are absent. Keep the wording fixed: change it every run and you end up measuring your own phrasing. We don't put numbers on what to expect, because we have no sourced figure to give. The method, and what to do with the results, is in [how to measure AI visibility](/how-to-measure-ai-visibility/). ## What to skip or treat with caution - **Guaranteed inclusion.** None of the platform documents we read offers one. - **Schema or special files sold as a Google requirement.** Google says there are no additional technical requirements beyond indexing and snippet eligibility. - **llms.txt as a proven lever.** It is a proposal, and the major AI crawler documents we read do not mention it. - **Keyword stuffing.** In the GEO paper's benchmark it did not perform well. - **robots.txt as protection.** RFC 9309 says its rules are not access authorisation, and some user-initiated fetchers may not follow them. - **The GEO paper as proof for your platform.** It is a benchmark study with stated conditions. - **Items with no stated basis.** If a checklist item can't say what it rests on, you can't weigh it. ## A short order of work | Step | What to do | Label | |---|---|---| | 1 | Check robots.txt and CDN rules for each AI search crawler | Documented | | 2 | Confirm pages are indexed and snippet-eligible in Google and Bing | Documented | | 3 | Make each page easy to quote: text, answer first, sourced and dated facts | Mixed | | 4 | Keep your name, place, offer and business profiles consistent and current | Mostly recommendation | | 5 | Add schema or llms.txt only if you have your own reason | Optional | | 6 | Re-run a fixed question set on a schedule | Mostly recommendation | Steps one and two carry the strongest evidence, so do them first. For the wider plan these steps sit inside, the home guide has a [30-day getting-started plan](/guide/#getting-started). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once the checklist is done, VisibilityToAI measures whether ChatGPT, Claude, Gemini and Perplexity actually name and cite you for your buyers' questions. - [Hilyt](https://hilyt.it/): For the entity step, Hilyt publishes an open AI-readable profile so the same name, place and offer are stated in one consistent place. ## Questions people ask ### What should be on a generative engine optimization checklist? Crawler access, indexing and snippet eligibility, quotable content, a clear and consistent business identity, and a fixed way to measure. The first two rest on platform documentation; the rest ranges from one research paper to our own recommendation. ### Is a GEO checklist different from an SEO checklist? Mostly it overlaps. Google says existing SEO fundamentals still apply to its AI features, and Bing says Copilot relies on the same crawling and indexing foundation as search. The additions are checking AI-specific crawlers and measuring what AI answers say rather than only rankings. ### Is there an AEO checklist, and is it the same thing? AEO, LLM SEO and GEO checklists describe almost the same work, so one list covers all three. The labels differ more than the steps do. ### Do I need llms.txt or schema markup on the checklist? Neither is required by Google, which says there are no additional technical requirements beyond being indexed and snippet-eligible. If you use structured data, Google says it should match the visible text, and llms.txt is a proposal that the major AI crawler documents we read do not mention. ### How do I know if the checklist worked? Put a fixed set of buyer questions to the same AI platforms on a regular schedule and record who is named, cited and how you are described. For Google, AI feature traffic is counted inside the Search Console Performance report's Web search type, not shown separately. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central: Creating helpful, reliable, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Bing Webmaster Tools: Webmaster Guidelines](https://www.bing.com/webmasters/help/webmaster-guidelines-30fba23a) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [llmstxt.org: The /llms.txt file, v2](https://llmstxt.org/) --- # Guide: Generative engine optimisation strategy: how to plan one, and what to decide first Canonical: https://whatisgeo.app/generative-engine-optimization-strategy/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > A GEO strategy is a short set of decisions: which buyer questions matter, which AI platforms your buyers use, what to fix first and how to check progress. Nobody can guarantee inclusion, so plan to iterate. Write the decisions down before you touch a single page: a goal, a fixed set of buyer questions, the platforms you will check, what you will fix first and how you will review it. The rest of this page takes those decisions in order. Each point is tagged. **Documented** means a platform or standard publishes it. **Research** means a study reported it under its own test conditions. **Recommendation** means it is our advice on sensible practice, with no official backing. ## What a GEO strategy is, and what it is not A strategy is the set of choices that tells you which tasks to do first. You decide what you want, where you will look for it and how you will know, and only then start editing pages. It is no list of tricks and no promise of visibility. Many agency pages rank "factors" and attach confident percentages to them without a source you can check. Nobody outside the platforms controls what an AI answer says, and answers can differ each time a question is asked. One documented anchor makes planning easier. Google says that to be eligible as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet, and that there are no additional technical requirements. It adds that specific optimisation is not required and that existing SEO fundamentals continue to be worthwhile (documented). So a GEO strategy builds on your SEO foundations; [GEO vs SEO](/geo-vs-seo/) sets out what overlaps. When you are ready for the doing, [how to improve AI visibility](/how-to-improve-ai-visibility/) covers the tactics. This page stays with the deciding. ## Start with the buyer questions, not the platforms Most people start by asking which platform to be on. Start instead with what your buyers ask: real questions, in their words, at the points where they compare options, check prices or look for someone nearby. Sales calls, support emails and your Search Console queries are good raw material. Our recommendation: write the list down and keep it fixed between checks, so that results from one round can be compared with the next. Group the questions by intent if it helps, for example choosing a provider, comparing alternatives and solving a specific problem. Keep it short enough that you will actually run it. You will revise it at reviews, because first lists are rarely right. The list shapes everything below. It tells you which pages need work, which facts about your business must be stated consistently, and where you are missing from answers altogether. ## Decide which AI platforms matter to your buyers Pick the platforms your buyers use, which may differ from the ones that get the most press. Ask customers how they research, check your analytics for referrals from AI tools, and think about your sector: a software buyer and someone hiring a local plumber may behave quite differently. If you are unsure, run your question list across the main assistants and let the results guide you (recommendation). For Google's AI features there is no separate report to plan around. Google says sites appearing in AI Overviews and AI Mode are included in the overall search traffic in Search Console, reported in the Performance report within the "Web" search type (documented). Name a primary few platforms in your plan and go deep on those, then revisit the choice at each review. ## Make sure you can be reached: access and indexing If a system cannot fetch your pages, little else matters, so settle access early. Vendors run separate crawlers for separate jobs, and each publishes its own list (documented): | Vendor | Search or indexing | Training | User-requested fetches | |---|---|---|---| | OpenAI | OAI-SearchBot | GPTBot | ChatGPT-User | | Perplexity | PerplexityBot | None listed; Perplexity says PerplexityBot is not used for foundation model training | Perplexity-User | | Anthropic | Claude-SearchBot | ClaudeBot | Claude-User | The details matter for strategy. OpenAI says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links, and that each of its settings is independent, so you can allow OAI-SearchBot while disallowing GPTBot. It also says robots.txt rules may not apply to ChatGPT-User, because those visits are started by a user. Perplexity says Perplexity-User generally ignores robots.txt for the same reason. Anthropic says its bots honour robots.txt and that disabling Claude-SearchBot or Claude-User may reduce your visibility in its search results. Check each vendor's page before you act, because names and descriptions change. The decision is which of these you allow, made on purpose. Blocking a training crawler and blocking a search crawler have different consequences. [AI crawlers and robots.txt](/ai-crawlers-robots-txt/) walks through the rules. Remember what robots.txt can and cannot do. RFC 9309, the standard behind it, describes rules that crawlers "are requested to honor" and says they "are not a form of access authorization" (documented). Treat the file as a statement of your wishes and never as a security control. On the Google side, Google lists checking that crawling is allowed in robots.txt and by any CDN or hosting infrastructure among its SEO fundamentals (documented). A firewall rule can shut a crawler out as surely as robots.txt can. ## Choose where to put effort: content, entity and third-party mentions You cannot do everything at once, so split effort across three places and decide the order: - **Your own content:** the pages you control, starting with those that answer your highest-value buyer questions. - **Your entity information:** the facts about who you are, what you do and where, stated the same way on your site, your profiles and your structured data. Google's list of SEO fundamentals includes keeping Business Profile and Merchant Center information up to date and making sure structured data matches the visible text on the page (documented). - **Third-party mentions:** what other sites, directories and reviews say about you. The platforms do not document how much this affects their answers, so treat it as a recommendation. **Research.** Aggarwal et al., "GEO: Generative Engine Optimization" (KDD 2024), tested content changes on GEO-bench, a benchmark of 10,000 queries, using a generative engine they built from the top five Google results plus a language model, and also ran tests on Perplexity.ai. Adding citations, quotations and statistics performed best on their visibility measures, with the abstract reporting gains of up to 40%, while keyword stuffing did poorly, and the effect varied across domains. These are the paper's own metrics in a test setup. Read the study as a sign that clear, well-sourced content helped in that setting; it does not forecast results for your site. ## A strategy by business type and budget Different businesses should lean different ways. This is our recommendation, with no statistics behind it. | Business type | Lean towards | Why | |---|---|---| | Local or service firm | Accurate entity information and reviews | Your name, services, area and contact details need to read the same everywhere | | Publisher | Crawler access and distinctive content | Your access choices decide who can use your pages, and original material is harder to replace | | SaaS or software | Comparison and documentation pages | Buyers ask comparison and how-to questions, and clear docs give answers something solid to cite | Budget sets the depth while the order stays the same. With little time: write the question list, check access and indexing, and fix one or two key pages. With more resource: widen the question set, cover more platforms and work on third-party presence. ## How to measure and review the strategy Decide the method before you start, so you are not inventing it afterwards to fit the results. Our recommendation: run the fixed question list on each chosen platform on a set schedule, record whether you are named, whether you are cited and what is said about you, and compare patterns across runs. Because answers vary, one result tells you little. Pair that with what you already have, including Search Console's Web search data and AI referrals in your analytics. [How to measure AI visibility](/how-to-measure-ai-visibility/) gives a full method, and an [AI visibility audit](/ai-visibility-audit/) gives you a baseline to measure from. Set a review rhythm you can keep. The point of a review is to change the plan when the evidence says so: drop a platform, add a question, or move effort to a different page. If reviews never change anything, they are not doing their job. ## What to leave out of a GEO strategy - **Guarantees.** No platform document we have read promises inclusion in an AI answer, so do not promise it to yourself, your boss or a client. - **Unsourced numbers.** If a "factor" comes with a precise percentage and no study you can read, skip it. The same goes for any timeframe nobody can back. - **Single screenshots.** One answer naming your brand is pleasant to see and proves little. - **Every tactic.** A strategy that lists every task becomes a checklist, and a checklist cannot tell you what to do first. ## A one-page strategy template If it does not fit on one page, it is too long. Write down: 1. **Goal.** What you want from AI answers, in one sentence. 2. **Question set.** Your fixed list of buyer questions, grouped by intent. 3. **Platforms.** The few you will check, and why your buyers use them. 4. **Access.** Which crawlers you allow, and confirmation that key pages are indexed and reachable. 5. **Effort.** The pages, entity facts and third-party sources you will work on first. 6. **Measurement.** How and how often you will run the questions, and what you will record. 7. **Review rhythm.** When you will review, and what changes you are prepared to make. If you are starting from nothing, the 30-day plan in [getting started](/guide/#getting-started) turns this template into a first month of work. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you would rather not rerun your question list by hand, it tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions. - [Hilyt](https://hilyt.it/): If your strategy puts entity information first, it gives you an open, AI-readable profile that sets out your business details in one place. ## Questions people ask ### What is a generative engine optimisation strategy? It is a short set of written decisions: which buyer questions matter, which AI platforms your buyers use, what you will fix first and how you will check progress. It comes before the tasks, and it cannot guarantee inclusion in AI answers because nobody can. ### What is the best GEO strategy for a small business? No single plan suits every business, but our recommendation is to keep it small: a short list of real buyer questions, a check that crawlers can reach your site, and accurate, consistent business details. Google's own SEO fundamentals include keeping your Business Profile up to date, which matters most for local firms. ### Is a GEO strategy different from an SEO strategy? They overlap heavily. Google says its AI features have no additional technical requirements beyond being indexed and eligible for a snippet, and that SEO fundamentals still apply. A GEO strategy adds decisions about AI crawler access and a way of checking whether AI answers name and cite you. ### How long does a GEO strategy take to show results? We cannot give a reliable timeframe, and no platform documentation we have read gives one. Answers vary between runs, so you need repeated checks over time to see a pattern. ### How often should you review a GEO strategy? Set a regular rhythm you can keep; how often depends on your resources and how quickly you can act on what you find. Each review should be able to change something: the questions, the platforms or where effort goes. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full text (HTML)](https://arxiv.org/html/2311.09735v3) --- # Guide: Generative engine optimization statistics: which numbers are real, and which to ignore Canonical: https://whatisgeo.app/generative-engine-optimization-statistics/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Few GEO statistics trace back to a study you can check. Platform docs give rules and almost no numbers, a few named studies hold up within their limits, and most headline percentages are vendor or consultancy claims. Only a handful of the numbers quoted about GEO lead back to something you can open and check. Even those describe one benchmark, one browsing panel or one survey at one point in time, so use them as context for your own measurements and never as a forecast for your site. Nobody can guarantee that a page will be included in an AI answer, and no statistic changes that. If you are new to the topic, start with [what GEO is](/guide/#what-is-geo) and [how it differs from SEO](/guide/#geo-vs-seo). ## How to grade a GEO statistic before you quote it Ask five questions of any figure. Who published it? Can you open the study itself? What was the sample, and who was in it? What date, platform and country does it cover? What do the authors say it cannot show? If you cannot answer the first two, stop there. A round percentage with no named study is a claim. It might be true, but neither you nor your reader can check it. Then put it in one of three grades: | Grade | What it is | What it can support | What it cannot | |---|---|---|---| | Platform documentation | The company that runs the product describes how it works | How the system behaves and how it is reported | Outcomes for any particular site | | Named research | A paper or analysis with a method you can read | What happened under the stated conditions | Other countries, periods, engines or sites | | Vendor or consultancy claim | A figure from someone who sells services, tools or advice | That the publisher says it | Anything, until the method is shown | ## What platforms themselves document Google's page "AI features and your website" is the clearest platform statement. To be shown as a supporting link in AI Overviews or AI Mode, "a page must be indexed and eligible to be shown in Google Search with a snippet", and it adds: "There are no additional technical requirements." It also says "specific optimization isn't required for AI Overviews and AI Mode", while normal SEO fundamentals "continue to be worthwhile". On measurement, the same page says sites appearing in AI features "are included in the overall search traffic" in Search Console, reported in the Performance report under the "Web" search type. Search Console's help page adds the counting rules: - Clicking a link to an external page in an AI Overview or AI Mode counts as a click. - An AI Overview occupies a single position, and every link in it is given that same position. - A follow-up question in AI Mode is treated as a new query. - Data from Search Labs experiments is not included. Neither page describes a separate report or filter for AI feature traffic, so you cannot read an "AI Overviews share" straight out of Search Console from what Google documents. One line on Google's page deserves a grade of its own. Google says that clicks from results pages with AI Overviews "are higher quality (meaning, users are more likely to spend more time on the site)". It publishes no figure or method for this, so treat it as the platform's own claim about its own product. We found no number in OpenAI, Perplexity or Anthropic crawler documentation that bears on visibility statistics, so they do not appear here. ## What named research has measured ### The "up to 40%" paper Aggarwal et al., "GEO: Generative Engine Optimization" (accepted to KDD 2024), is where the term and the most quoted number come from. What the paper actually did: - **Benchmark.** GEO-bench, 10,000 queries drawn from nine sources, about 80% of them informational. - **Engine.** For each query, the authors fetched the top five Google results and had gpt-3.5-turbo write an answer citing them, sampling five answers per query. - **Methods.** Nine content rewrites, including Cite Sources, Quotation Addition, Statistics Addition, Fluency Optimization and Keyword Stuffing. - **Metrics.** Position-Adjusted Word Count (how much of the answer is attributed to a source, weighted by where its citation appears) and Subjective Impression (seven factors scored by GPT-3.5). The best methods improved on the unchanged baseline by 41% on Position-Adjusted Word Count and 28% on Subjective Impression. The authors summarise this as "up to 40%". Keyword stuffing gave little to no improvement, and results varied by domain. They also tested the methods on Perplexity.ai and report improvements of up to 37% there. The limits matter more than the headline. The gain is a relative change in one source's share of an answer built from five sources, measured mostly on an engine the authors built themselves with a 2023 to 2024 model. It does not measure traffic, rankings, leads or sales. Quote it as "a research benchmark reported relative visibility gains of up to 40% under its own conditions". ### Pew Research Center on clicks Pew's 2025 analysis is independent research with a published method. It used the browsing data of 900 US adults in its KnowledgePanel Digital panel, who agreed to have their browsing tracked, from 1 to 31 March 2025. That gave 68,879 unique Google searches, of which 12,593 showed an AI summary. Pew re-ran the searches with a scraping service between 7 and 17 April 2025 to see what each results page contained. What it found: - Users clicked a traditional search result on 8% of visits to pages with an AI summary, against 15% of visits without one. - They clicked a link inside the AI summary on 1% of visits to pages that had one. - They ended their browsing session on 26% of pages with an AI summary, against 16% without. - 18% of the Google searches in the study produced an AI summary. Its limits: Google only, US adults only, one month, and results pages were collected a few weeks after the searches were made, so some summaries may have changed. A "click" was inferred from the next URL visited. ## Why vendor and roundup numbers need care A company that sells GEO services, tools or advice has a reason to publish a big number. That does not make the number false, but it comes with an interest attached and the method is often missing. Roundups make it worse. One page quotes a percentage, the next quotes the first page, and soon a figure appears on dozens of sites with the original sample, date and wording stripped away. By then "80% of search users rely on AI summaries at least 40% of the time" has often become "80% of people use AI to search". ## Common headline claims and what is behind them We traced the four figures that turn up most often in the current top results. | Claim as usually quoted | Who actually published it | What we could check | Grade | |---|---|---|---| | "80% rely on AI summaries" | Bain & Company press release, 19 February 2025: "About 80% of search users rely on AI summaries at least 40% of the time" | The press release gives no sample size or method on the page. It also claims this reduces organic web traffic "by 15% to 25%" | Consultancy claim | | "60% of searches end without a click" | The same Bain release: "On traditional search engines, about 60% of searches now end without the user progressing to another destination" | No method on the page | Consultancy claim | | "Organic traffic down 50% by 2028" | A Gartner Insights abstract (5 May 2026): "Gartner predicts organic search traffic will decrease by 50% by 2028" | It is a forecast, and the research behind it is paywalled, so we could not read the method or assumptions | Analyst forecast | | "37% start searches with AI" | Eight Oh Two, an SEO and PPC agency, as reported by Search Engine Land | A survey of 500 consumers who actively use AI tools, data collected in November. The full report requires registration, so we read the coverage only | Agency claim | Two points stand out. The 37% describes people who already use AI tools, so it says nothing about consumers in general. And it has nothing to do with the "up to 37%" Perplexity.ai result in the Aggarwal paper; the two numbers simply share a value. None of this makes the figures wrong. It means each one is the publisher's own claim, and should be quoted with that publisher's name attached. ## What you can measure for your own site This section is our recommendation, not a research finding. Your own numbers will tell you more than anyone's headline, because they describe your pages and your buyers. 1. **Search Console for Google.** Watch clicks and impressions in the Performance report. Remember that AI Overview and AI Mode activity is counted inside the "Web" search type rather than reported separately. 2. **Referrals in your analytics.** Look for visits referred by AI assistants and note how they change over time. 3. **A fixed set of buyer questions, checked on a schedule.** Write down the questions your customers actually ask, run them on the platforms you care about, and log whether your business is named, cited and described accurately. Keep the question set fixed so the comparison means something, and record the date, platform, location and exact wording each time. Do the same for any outside statistic you quote: write down its sample size, date, platform, prompt set and location. More on setting this up is in the [measurement section](/guide/#measurement) of our guide and in [what AI visibility means](/what-is-ai-visibility/). For Google specifically, see [how to appear in AI Overviews](/how-to-rank-in-google-ai-overviews/). ## How to use statistics in a business case without overclaiming Lead with what you measured yourself. Add outside figures as context, each labelled by grade. Some wording that stays accurate: - "A KDD 2024 research benchmark reported relative visibility gains of up to 40% for some content changes, under its own test conditions." - "Pew Research Center found that in March 2025, US adults in its panel clicked a traditional result on 8% of Google visits with an AI summary, against 15% without." - "Bain & Company says about 80% of search users rely on AI summaries at least 40% of the time." Never turn any of these into a promise. A sound business case says what you will track, how often, and what you will do if the numbers do not move. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you would rather track your own numbers than quote someone else's, it checks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions. ## Questions people ask ### Where do the statistics on GEO and AI search come from? Three places: platform documentation (which gives rules and very few numbers), named research such as the Aggarwal et al. GEO paper and Pew Research Center's browsing study, and vendor, agency or consultancy reports. Only the first two usually show a method you can read. ### Is it true that GEO can boost visibility by up to 40%? It is a reported research result with narrow conditions. Aggarwal et al. measured relative gains in a word-count-and-position metric for one source inside answers generated by GPT-3.5-turbo from the top five Google results on their GEO-bench benchmark. It does not measure traffic, rankings or sales, and it is not a guarantee for any live product. ### Do AI Overviews reduce clicks to websites, and what is the evidence? Pew Research Center tracked the browsing of 900 US adults in March 2025 and found they clicked a traditional search result on 8% of Google visits that showed an AI summary, against 15% of visits without one. It covers Google only, in the US, for one month. ### Can I trust the percentages in vendor GEO reports? Treat them as the publisher's own claims. Before quoting one, find the sample size, who was sampled, the date, the platform and the method. If you cannot find those, use it for ideas and leave it out of your evidence. ### How do I measure my own GEO numbers instead of quoting others? Track Google search performance in Search Console, watch referrals from AI assistants in your analytics, and check a fixed set of buyer questions on each platform on a regular schedule. Log the date, platform, location and exact wording every time so the results stay comparable. ## Sources - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Search Console Help: how clicks, impressions and position are counted (AI Overviews and AI Mode)](https://support.google.com/webmasters/answer/7042828) - [Pew Research Center: Google users are less likely to click on links when an AI summary appears (22 July 2025)](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) - [Bain & Company press release: consumer reliance on AI search results (19 February 2025)](https://www.bain.com/about/media-center/press-releases/20252/consumer-reliance-on-ai-search-results-signals-new-era-of-marketing--bain--company-about-80-of-search-users-rely-on-ai-summaries-at-least-40-of-the-time-on-traditional-search-engines-about-60-of-searches-now-end-without-the-user-progressing-to-a/) - [Gartner Insights abstract: Hire an AEO Manager to Succeed in the Zero-Click Era (5 May 2026)](https://www.gartner.com/en/documents/7807617) - [Search Engine Land: 37% of consumers start searches with AI instead of Google (coverage of Eight Oh Two survey)](https://searchengineland.com/consumers-start-searches-ai-not-google-study-467159) --- # Guide: Generative engine optimization tools: what they do and how to choose one Canonical: https://whatisgeo.app/generative-engine-optimization-tools/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > GEO tools track whether AI answers mention or cite you, audit your site for crawler access, or help with content. Pick by the job you need done, and treat any promise of guaranteed AI inclusion as a red flag. Before you compare products, work out which problem you have. You may not know whether AI answers mention you, you may suspect crawlers can't reach your pages, or you may need help with content. Each one needs a different kind of tool, and the platforms already give you some of the data free. Most "best GEO tools" pages are written by vendors. They put their own product first and don't say how they scored anything. This guide sorts tools by the job they do and gives you questions to ask before you pay. ## What GEO tools do, and what none of them can do A tracker sends questions (prompts) to AI engines and records the answers. An audit tool looks at your site from the outside: robots.txt rules, whether pages can be reached and how the content is structured. A content assistant suggests edits. All three can be useful, but they only observe and advise. None of them controls what an engine writes. Google's documentation is plain about its own AI features. To be eligible as a supporting link in AI Overviews or AI Mode, "a page must be indexed and eligible to be shown in Google Search with a snippet", and "there are no additional technical requirements". It adds that "specific optimization isn't required" and that the usual SEO basics still apply. Be wary of any tool that promises guaranteed inclusion in AI answers, because Google documents no special lever for it. For how this fits with ordinary search work, see [GEO vs SEO](/guide/#geo-vs-seo). ## The four kinds of tool | Job | What it tells you | What it can't tell you | |---|---|---| | Visibility and citation tracking | Whether sampled AI answers name or link to you | What every user sees, or why an engine chose a source | | Site and crawler audit | Whether bots are allowed in and can read your pages | Whether you will be cited once they can | | Content assistance | Suggested edits and structure | Whether those edits will change any engine's answer | | Free first-party reports | What the platform itself logged about your site | Competitors, or questions you never appeared for | Many products bundle two or more of these and call the result a platform. That is fine, but judge each part on its own job. ### Visibility and citation trackers These run a set of prompts against one or more AI engines. They log whether your brand is named, whether your pages are linked as sources, and sometimes the tone of the answer. Keep mentions and citations apart. A mention means your name appears in the answer text. A citation means one of your URLs appears as a source. You can get one without the other, and a dashboard that merges them hides which problem you have. [What is AI visibility?](/what-is-ai-visibility/) explains the difference in more detail. ### Site and crawler audits An audit checks whether AI crawlers can reach you at all. A useful one compares your robots.txt with the user agents each operator publishes: - **OpenAI:** OAI-SearchBot (ChatGPT search results), GPTBot (model training) and ChatGPT-User (fetches made on a user's behalf; OpenAI says robots.txt rules may not apply to these). - **Anthropic:** ClaudeBot (training), Claude-SearchBot (search quality) and Claude-User (fetches when someone asks Claude a question). - **Perplexity:** PerplexityBot (search results) and Perplexity-User (user-requested fetches, which Perplexity says generally ignore robots.txt). - **Google:** Google says robots.txt rules for Googlebot are how you control crawling for Search, and AI Overviews and AI Mode are part of Search. Google-Extended is a separate robots.txt token for Gemini training and grounding, and Google says it does not affect inclusion in Google Search. That last point is a handy test of an audit tool. If it tells you to allow Google-Extended so you can appear in AI Overviews, it has misread Google's documentation. Operators change these lists, so check each operator's page rather than relying on any copy, including this one. The robots.txt standard, RFC 9309, describes rules that crawlers "are requested to honor" and states that "these rules are not a form of access authorization." robots.txt asks crawlers to stay out; it doesn't lock anything. The practical steps are in [crawler access](/guide/#crawler-access). ### Content and optimisation assistants These suggest edits, such as putting the answer near the top, tightening claims and improving structure. Some score a draft against a target query. Be careful with the scores. The research most often cited here is Aggarwal et al., *GEO: Generative Engine Optimization* (KDD 2024). It tested content changes on its own benchmark, GEO-bench, with its own generative engine setup, and found that how well a method worked varied by domain. It is a useful study, but it does not show that the same changes will work on ChatGPT or Google today. When a vendor says a tactic raises AI visibility by some percentage, ask which engine, which prompts and how it was measured. Clear, accurate writing helps readers whatever an engine does, so judge these tools on whether their edits make the page better. ### Free first-party reports Before paying anyone, look at what the platforms already give you: - **Google Search Console.** Google says traffic from AI Overviews and AI Mode is included in the overall search traffic in the Performance report, under the "Web" search type. Its help pages explain how it is counted: a click on an external link in either feature counts as a click, and all the links in an AI Overview share one position. - **Bing Webmaster Tools.** Microsoft announced an AI Performance report in public preview in February 2026. It shows how often your pages are cited as sources in Microsoft Copilot, in AI-generated summaries in Bing and in some partner integrations, which URLs are cited, and a sample of the "grounding queries" used to find them. These reports come from the platforms' own data, which is their strength. They also have limits. Each covers only its own platform, shows nothing about competitors and leaves out questions you never appeared for. ## What to ask any vendor before you pay These questions are our own recommendation. The answers will tell you more than the demo. 1. **What is the prompt set?** Who chose the prompts, how many are there, and can you add your own? Generic prompts only tell you about generic prompts. 2. **How many repeat runs per prompt?** One run is an anecdote. Ask how many samples sit behind each number. 3. **Which engines, exactly?** Ask for the specific products, such as ChatGPT search or Google AI Mode. "AI search" is too vague. 4. **Logged-out interface or API?** Does the tool collect answers as a logged-out user sees them, or through an API that may behave differently? Either can be valid, provided you know which one you are buying. 5. **Citation or mention?** Are they reported separately? 6. **Can you export the raw data?** You want rows with the date, engine, prompt and full answer text as well as the charts. 7. **What are the contract and cancellation terms?** Read them before you sign. Then compare every claim the vendor makes with how the tool actually collects answers. If a claim doesn't fit the collection method, treat it as marketing. ## Why tool numbers are samples of answers AI answers are not fixed. Ask the same question twice and you can get different brands and different sources. Google also says AI Mode and AI Overviews may use different models and techniques, so the responses and links they show will vary. That makes a tracker's output a sample. The tool has no fixed list to read from. When a dashboard puts you "third", it means that, in the answers it collected with its prompts on those dates, you appeared in that pattern. Another run can change it. Sampling still has value. A consistent trend across many samples tells you something, and one reading tells you very little. Give more weight to the direction over time than to any single figure. ## Free checks to run before buying anything This is the order we suggest. It costs nothing but time. 1. **Write a fixed prompt list.** Use questions your customers really ask, and keep the wording the same between runs. 2. **Run each prompt more than once** on the engines you care about, logged out where you can. Record mentions and citations separately. 3. **Keep your own log.** A spreadsheet is enough: date, engine, prompt, full answer, named or not, linked or not. 4. **Check robots.txt** against the user agents above, and check that your CDN or host isn't blocking the bots you want to let in. 5. **Check indexing** with Search Console's URL Inspection tool, because Google's requirement is that a page is indexed and eligible for a snippet. 6. **Read the first-party reports** in Search Console and Bing Webmaster Tools. After a few rounds you will usually know whether your problem is measurement, access or content. That tells you which kind of tool, if any, is worth paying for. For the wider first month, see [getting started](/guide/#getting-started). For how to read your log over time, see [measuring progress](/guide/#measurement). ## Which setup fits which kind of business These are our recommendations: - **A small business with a few key pages:** the free checks and a spreadsheet are usually enough, because the log stays small enough to read yourself. - **A larger site with many products or markets:** manual runs stop being practical, so this is where a tracker earns its place. Pair it with a regular crawler audit. - **An agency with several clients:** put export, a fixed prompt set per client and open methods ahead of extra features. - **A publisher or content-heavy site:** start with the audit and content jobs, because access and structure come before anything else. Whichever setup you choose, the habits in [our principles](/guide/#principles) still apply. ## How we handle this page We have not ranked any tools, and there is no "best" list here. A fair ranking would need a stated, repeatable test across products. Without one, it would be exactly the kind of unexplained verdict this page warns about. We quote no prices either, because we have not checked any vendor's pricing page for this guide. If we add vendor figures later, each will come from the vendor's own page, credited to the vendor and dated. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you outgrow a manual prompt log, it is one option for tracking whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions; put the same vendor questions to it as to any other tool. ## Questions people ask ### What are generative engine optimization tools? They are software that tracks, audits or helps improve how your business appears in AI-generated answers. Most do one of three jobs: tracking mentions and citations in sampled answers, auditing crawler access and content, or suggesting content edits. ### Do I need a paid tool to track my AI visibility? Not to start. A fixed prompt list, repeat runs and a spreadsheet log, plus Google Search Console and Bing Webmaster Tools, will tell you a lot. A paid tracker earns its place when there are too many prompts to run by hand. ### How accurate are AI visibility tracking tools? A tracker can only be as reliable as its sampling. AI answers vary between runs, so a tool's numbers are samples of answers and should not be read as fixed rankings. Ask about the prompt set, repeat runs and collection method, and read trends over time rather than single figures. ### What is the difference between GEO tools, AEO tools and LLM SEO tools? In our view the labels overlap so much that they are mostly branding, and vendors use them interchangeably. Ignore the label and look at the job the tool does and how it collects its data. ### Can any tool guarantee my site is cited by ChatGPT or Google AI Overviews? No. No tool controls what an engine writes. Google documents that AI Overviews and AI Mode have no extra technical requirements beyond being indexed and eligible for a snippet, and that specific optimisation isn't required. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Search Console Help: Performance report (Search results)](https://support.google.com/webmasters/answer/7576553) - [Search Console Help: How impressions, position and clicks are counted (AI Overviews, AI Mode)](https://support.google.com/webmasters/answer/7042828) - [Bing Webmaster Blog: Introducing AI Performance in Bing Webmaster Tools Public Preview](https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Google: List of Google's common crawlers (Google-Extended)](https://developers.google.com/search/docs/crawling-indexing/google-common-crawlers) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: GEO vs SEO: what is actually different, and what is the same? Canonical: https://whatisgeo.app/geo-vs-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > GEO and SEO share most of their foundations: Google and Bing both say SEO basics are what make pages eligible for their AI answers. What changes is the goal (being cited, not only ranked) and how you measure it. Short answer: if a page cannot be crawled, indexed and understood, it will struggle in both classic search and AI answers, so the SEO basics come first. GEO adds a different target on top of them. You want to be named, quoted or described accurately inside an answer, and you need a different way to check whether that is happening. Most pages on this topic blend three kinds of claim. Here each one is labelled: **documented** (a platform's own documentation says so), **research** (a named study, with its limits) or **our recommendation** (sensible practice, nothing more). For the wider background, start with [what GEO is](/guide/#what-is-geo). ## The short answer: mostly the same foundations, a different goal Same plumbing, different target. A page still has to be reachable by crawlers, indexed and useful. What changes is what success looks like. In classic search you want a position on a results page and a click. In a generative answer you want to be one of the sources the engine draws on, ideally with the facts about your business stated correctly, even if the reader never clicks. ## What SEO aims at, and what GEO aims at SEO aims at ranking. Your page competes with others for a query, the engine shows a list, and the reader chooses. GEO aims at inclusion. A generative engine takes a question, gathers sources and writes an answer. Bing's guidelines describe GEO as focusing "on content eligibility for grounding and reference in AI responses" (documented). In practice the engine does more of the choosing, and the reader may never see the full list of sources. ## GEO vs SEO side by side | | SEO | GEO | |---|---|---| | Main goal | Rank for a query | Be cited or described in an answer | | What success looks like | A position, a click, a conversion | A mention, a citation, an accurate description | | Technical foundations | Crawlable, indexed, useful pages | The same | | Content emphasis | Matching intent, depth, structure | The same, plus clear facts that are easy to quote | | Crawler decisions | Mainly search crawlers | Search, training and user-triggered agents, controlled separately | | Measurement | Rankings, traffic, conversions | Asking buyer questions and checking the answers, plus search data | The middle rows barely change. The differences sit at the top and bottom: what you want, and how you check whether you got it. For how an AI answer is put together, see [how AI answers happen](/guide/#how-it-works). ## What each platform says about eligibility Everything in this section is documented behaviour. Where a platform says nothing, we leave the gap. **Google (AI Overviews and AI Mode).** To be shown as a supporting link, "a page must be indexed and eligible to be shown in Google Search with a snippet", and "there are no additional technical requirements." Google adds that specific optimisation "isn't required" for these features, while existing SEO fundamentals "continue to be worthwhile". It also notes that the two features may use different models and techniques, so the links they show will vary. **Bing (Copilot).** Bing's guidelines say its search and Copilot experiences "rely on the same core crawling, indexing, and ranking foundation as traditional search", and that SEO best practices also support eligibility for AI answers and citations. They also flag that some page controls affect Copilot: for example, NOCACHE limits Copilot to the URL, title and snippet. **OpenAI (ChatGPT).** OAI-SearchBot surfaces sites in ChatGPT's search features, while GPTBot crawls content that may be used to train its models. Each robots.txt setting is independent, so you can allow one and block the other. OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. **Perplexity.** PerplexityBot surfaces and links websites in Perplexity's search results and, according to Perplexity, is not used to crawl content for AI foundation models. Perplexity-User fetches pages when a user asks a question, and Perplexity says it "generally ignores robots.txt rules" because the user requested the fetch. **Anthropic (Claude).** Anthropic lists three bots. ClaudeBot collects content that may be used for model training. Claude-User fetches pages when people ask Claude questions. Claude-SearchBot indexes content to improve search results. Anthropic says its bots honour robots.txt, and that blocking Claude-User or Claude-SearchBot may reduce your visibility in Claude's search answers. **A caveat for all of them.** robots.txt is defined in RFC 9309, and the RFC is explicit that its rules "are not a form of access authorization." Crawlers are asked to follow them; nothing forces them to. Setup details are in [crawler access](/guide/#crawler-access). ## What the research says (and its limits) The main academic reference is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024 (research). The authors define GEO and build GEO-bench, a benchmark of 10,000 queries. Each query comes with the text of the top five Google results, and a language model writes an answer from those sources. In their tests, adding citations, quotations from relevant sources and statistics made a source more visible in the generated answers, and they report gains of up to 40% on their own visibility measures. Keyword stuffing, a classic SEO tactic, gave little or no improvement. They also ran the methods on Perplexity.ai. The limits matter. These results come from a benchmark the authors built, metrics they defined and engines as they behaved when tested. The paper itself says effectiveness varied by domain. Live AI products change their retrieval and answer style over time, so treat the findings as evidence of where to experiment. They are no guarantee of results on any live platform today. ## Where GEO and SEO overlap Mostly everywhere. Both need pages that can be crawled and indexed, content that answers the question properly, and a site that is technically sound and clear about who is behind it. Google's AI features page lists ordinary SEO work as still worthwhile: allowing crawling in robots.txt and at your CDN, internal links, page experience, important content in text form, and structured data that matches the visible text (documented). If your SEO is weak, GEO work will not make up for it. ## Where GEO adds work This section is our recommendation. It goes beyond documented platform requirements. 1. **Make key facts easy to quote.** State what you do, who for, where, and the details buyers ask about, in plain sentences a system can lift cleanly. 2. **Decide which crawlers you want.** OpenAI, Perplexity and Anthropic each separate search, training and user-triggered agents. Choose on purpose, then check your robots.txt and firewall rules match. 3. **Check how AI describes you.** Ask the questions a buyer would ask and read the answers for errors or omissions. 4. **Measure beyond rankings.** A ranking report will not tell you whether you appear in an answer. None of this is exotic. It is an extra layer on top of normal SEO. ## Does GEO replace SEO? No. Google says AI Overviews and AI Mode need nothing beyond being indexed and eligible for a snippet. Bing says Copilot runs on the same foundation as its search. OpenAI, Perplexity and Anthropic all point site owners to robots.txt, the same access layer SEO has always relied on. Nothing in the documentation we read suggests the foundations stopped mattering. What nobody can guarantee is a particular outcome. Bing puts it plainly: "SEO does not guarantee rankings or traffic, and GEO does not guarantee grounding or citations in AI experiences." Google likewise says it cannot guarantee that content will be crawled, indexed or served. If someone promises you a guaranteed place in AI answers, ask for their source. The home guide's [GEO vs SEO section](/guide/#geo-vs-seo) has the short version. ## GEO, AEO and 'search everywhere' optimisation: are they different? Mostly they are different labels for overlapping work. AEO (answer engine optimisation) is generally used to mean getting your content used as the answer to a question. "Search everywhere optimisation" is generally used to mean being findable on many platforms beyond one search engine. GEO is about generative answers specifically. No standards body defines any of these terms, so do not worry about which label someone prefers. Ask what they would change on your site. One related question comes up often: is Google Ads part of SEO? No. Ads are paid placements. SEO concerns unpaid results, and Google states it does not accept payment for crawling, indexing or ranking. ## How to measure each one SEO measurement is mature: rankings, organic traffic and conversions. Google says traffic from AI Overviews and AI Mode is counted within the "Web" search type in Search Console's Performance report, so your existing reports already include it (documented). Bing notes that content can appear as impressions, citations or grounding references in Copilot, and that "a decline in clicks does not always indicate a loss of visibility" (documented). For the AI side, our recommendation is to keep a fixed list of buyer questions, ask them in the AI tools your customers use, and record what comes back. Is your business named? Is it described correctly? Which sources are cited? Repeat this on a schedule, because answers vary between runs and change over time. More on this in [measuring progress](/guide/#measurement). ## What to do first This sequence is our recommendation: 1. Fix the SEO fundamentals: crawlable, indexed, snippet-eligible pages with useful content. 2. Write your key facts clearly, in text, where crawlers can reach them. 3. Decide which search, training and user agents to allow, and set robots.txt to match. 4. Ask AI tools your buyers' questions and note what they say about you. 5. Track those answers alongside your search data, without relying on rankings alone. For a fuller plan, see [getting started](/guide/#getting-started). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once the foundations are in place, it checks whether ChatGPT, Claude, Gemini and Perplexity actually name and cite your company for your buyers' questions. ## Questions people ask ### What is the difference between GEO and SEO? SEO aims to rank pages in search results. GEO aims to get your content cited or accurately described in AI-generated answers. The technical foundations are largely shared; the goal and the way you measure it differ. ### Is GEO replacing SEO? No. Google says pages need no additional technical requirements for AI Overviews and AI Mode beyond being indexed and eligible for a snippet, and Bing says Copilot relies on the same crawling, indexing and ranking foundation as its search. GEO builds on SEO. ### Do I need to do GEO if my SEO is already good? Good SEO covers most of what the platforms document. Our recommendation is to add a few things on top: make key facts easy to quote, choose which AI crawlers to allow, and regularly check how AI tools describe your business. ### What is AEO vs GEO vs SEO? SEO targets rankings in search results. AEO (answer engine optimisation) and GEO both target being used in answers rather than a list of links. No standards body defines these labels, so ask anyone using them what they would actually change. ### Does Google treat AI Overviews content differently from normal ranking? Google documents no extra requirements: a page must be indexed and eligible to show with a snippet, and existing SEO fundamentals still apply. Google also notes that AI Overviews and AI Mode may use different models and techniques, so the links they show vary. ### Is Google Ads part of SEO? No. Google Ads is paid advertising, while SEO concerns unpaid results. Google states that it does not accept payment for crawling, indexing or ranking. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Bing Webmaster Guidelines](https://www.bing.com/webmasters/help/webmaster-guidelines-30fba23a) - [Google Search Central: In-depth guide to how Google Search works](https://developers.google.com/search/docs/fundamentals/how-search-works) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization (full text, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: Google AI Mode and SEO: what changes, and what to do about it Canonical: https://whatisgeo.app/google-ai-mode-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Google says the same SEO foundations apply to AI Mode, with no extra technical requirements. A page must be indexed and eligible to show a snippet to be linked, and nobody can guarantee a citation. In practice, ordinary SEO done properly is what counts: AI Mode links to pages that are in Google's Search index. Three documented things deserve your attention: who is eligible, the snippet controls that decide what AI Mode may use, and how its traffic shows up in Search Console. Throughout, we say which claims come from Google's documentation, which come from research and which are our own recommendation. ## What Google AI Mode is, in one paragraph AI Mode is a conversational search experience inside Google Search. Google's help page describes it as an AI-powered response "with the ability to go deeper through follow-up questions and helpful links to the web", and says it expands what AI Overviews can do with more advanced reasoning. You reach it at google.com/ai, from the AI Mode option after typing a question on google.com, or in the Google app. When we checked on 18 October 2026, the help page listed a long run of supported countries and territories, including the United Kingdom and the United States, and dozens of languages, English among them. Google says it will keep refining the product, so check that page for the current list. ## What Google says you need to do (and what it says you do not) Google's page "AI features and your website" covers AI Overviews and AI Mode together. In its own words, you can apply "the same foundational SEO best practices for AI features as you do for Google Search overall", and "specific optimization isn't required for AI Overviews and AI Mode". On technical requirements it is blunt: "There are no additional technical requirements." It then lists SEO fundamentals it still calls worthwhile. This list is Google's: - allow crawling in robots.txt, and by any CDN or hosting infrastructure - make content easy to find through internal links - provide a good page experience - make sure important content is available in textual form - support text with high-quality images and video where relevant - make sure structured data matches the visible text on the page - keep Merchant Center and Business Profile information up to date That is the documented baseline. Many top results for this search come from agencies and vendors with a reason to say the rules have changed, and Google's own documentation does not say so. Documentation can't tell you everything about how a system behaves. Still, it is the only primary source on what Google asks of site owners, so start there. ## How AI Mode finds pages: query fan-out and what it means for your content Google says both AI Overviews and AI Mode "may use a 'query fan-out' technique", issuing multiple related searches across subtopics and data sources to build a response. The AI Mode help page puts it more plainly: AI Mode divides your question into subtopics, searches for each one at the same time, then brings the results together. Google says this lets it show "a wider and more diverse set of helpful links" than a classic web search. Google also says AI Mode and AI Overviews "may use different models and techniques", so the responses and links they show will vary. The guidance for site owners is the same for both, but you should expect different outputs. What this means for your content is our interpretation, because Google publishes no recipe. If one question turns into several searches, a page that covers the question properly, plus the subtopics and follow-ups a reader would naturally have, has more chances to match one of them. Padding is a different matter. Thin sections written to touch more phrases help nobody, and the GEO research below found keyword stuffing did little. For the bigger picture of how AI answers are put together, see [how AI answers happen](/guide/#how-it-works). For where this overlaps with ordinary search work, see [GEO and SEO](/guide/#geo-vs-seo). ## Eligibility and controls: indexing, snippets, nosnippet and robots This part is documented and mechanical, so it is worth getting exactly right. **Eligibility.** Google says: "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet". That is two conditions. If a page fails either one, it can't be a supporting link. **Controls.** Google says robots.txt rules for Googlebot are how you manage crawling for Search, AI features included. To limit what is shown, it points to `nosnippet`, `data-nosnippet`, `max-snippet` and `noindex`. Its robots meta tag documentation spells out what each one does to AI features: | Control | What Google documents | | --- | --- | | `noindex` | The page is not shown in search results, so it can't be a supporting link. It also drops out of ordinary results. | | `nosnippet` | No text snippet or video preview in any form of Google search result, which Google says includes AI Overviews and AI Mode. The content is also not used as a direct input for them. | | `max-snippet:[number]` | Caps snippet length and limits how much of the content may be used as a direct input for AI Overviews and AI Mode. `max-snippet:0` is the same as `nosnippet`. | | `data-nosnippet` | An HTML attribute for `span`, `div` or `section` elements. It keeps those parts of the page out of snippets. | Watch for two traps. First, Google can only read these rules on pages it is allowed to crawl. If robots.txt blocks a page, Google never sees its `noindex`. Second, Google-Extended. Google's crawler documentation says this token controls whether your content may be used to train future Gemini models and for grounding in Gemini Apps and Vertex AI. It also says Google-Extended "does not impact a site's inclusion in Google Search". Blocking it will not keep you out of AI Mode. The home guide's section on [crawler access](/guide/#crawler-access) covers crawlers more widely. Restricting snippets is a trade-off, because it limits what AI Mode can use and show. Choose it deliberately. What if you have added a preview control and your content still appears in AI features? Google suggests using the URL Inspection tool to check that Googlebot actually received the control, then allowing time for a recrawl. It says recrawling can take "several days to several months". ## What to check and improve: a practical checklist This checklist is our recommendation, built on Google's list above. None of it is a documented ranking requirement, and doing all of it guarantees nothing. 1. **Confirm your important pages are indexed.** Verify your site in Search Console and run URL Inspection on the pages that matter. 2. **Look for snippet blocks you didn't mean to add.** Search your templates for `nosnippet`, `max-snippet` values and stray `data-nosnippet` wrappers, and check HTTP headers for `X-Robots-Tag`. 3. **Check robots.txt and your CDN.** Make sure Googlebot isn't blocked by a rule, or by bot protection at the hosting layer. 4. **Keep the answer in text.** If key information lives only in an image, a PDF or a widget a crawler can't read, treat it as missing. 5. **Answer the next questions.** For each important page, list what a reader would ask next, and answer it on that page or a linked one. 6. **Keep titles and structured data accurate.** A title should describe the page honestly, and structured data should match what is visible. Nothing exotic, which matches Google's guidance. The home guide's [principles](/guide/#principles) explain the thinking behind most of it. ## What the research says, and its limits The best-known study is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024. The authors built GEO-bench, a set of 10,000 queries from several sources. They also built a test generative engine: for each query it fetched the top five Google results and had a language model write an answer from them. They then rewrote source pages in different ways and used metrics they defined themselves to measure how much each source was used and cited. Here is what they found under those conditions. Adding citations, quotations from relevant sources and statistics worked best, with relative improvements of roughly 30 to 40% on their position-adjusted word count metric. Keyword stuffing, a classic SEO tactic, gave "little to no improvement". They also tested the same methods on Perplexity.ai and report improvements there too. The limits matter. The study did not test Google AI Mode. Its main results come from a research engine run on a benchmark and scored with the authors' own metrics. The authors also note that results varied by domain. Treat the study as evidence that the way a page is written can change how generative engines use it. It does not tell you what will happen in AI Mode. The same caution applies to vendor claims. This page uses no traffic, click-through or ranking-factor numbers from agencies or tool vendors. ## How to see AI Mode in your reporting Google says traffic from AI features is included in your overall search traffic in Search Console, in the Performance report under the "Web" search type. That report has no separate AI Mode filter. Search Console's help page on how data is counted gives these details for AI Mode: - a click on a link to an external page in AI Mode counts as a click - standard impression rules apply - position is worked out the same way as on a normal results page - a follow-up question in AI Mode counts as a new query - Search Labs experiments are not included So there is no report where you can read off "AI Mode clicks". Our recommendation is to work around that: 1. Pick a set of pages that matter, and record their clicks, impressions and queries under the Web search type. 2. Keep a dated log of every change you make to those pages. 3. Compare later periods with that baseline, reading any changes against your log and against wider shifts in your traffic. This is a blunt method, but it is honest about what the data can show. The [measurement section](/guide/#measurement) of the home guide covers setting up a repeatable baseline, including checks on AI assistants outside Google. ## What you cannot control You can't guarantee a citation or force a link into a generated answer. You can't isolate AI Mode traffic in Search Console, and you can't see how fan-out splits a given question. Google itself says AI Mode "doesn't always get it right" and may misinterpret web content or miss context. These things you can control: whether your pages can be crawled and indexed, whether snippets are allowed, whether your content answers real questions in readable text, and whether your measurement is set up so you actually learn from it. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Search Console cannot show you whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions, and this measures that. ## Questions people ask ### Do I need to do anything different for Google AI Mode? Google says no. The same foundational SEO best practices apply and there are no additional technical requirements. A page has to be indexed and eligible to show with a snippet before it can be a supporting link. ### Is AI Mode SEO different from AI Overviews SEO? Google gives one set of guidance that covers both. It also says the two may use different models and techniques, so the answers and links each one shows will vary. ### How do I stop my content appearing in AI Mode? Use nosnippet, or max-snippet, to stop or limit your content being used as a direct input. Use data-nosnippet for parts of a page, or noindex to keep the page out of Search entirely. Blocking Google-Extended will not do it, because Google says it does not affect inclusion in Google Search. ### Will AI Mode reduce my traffic? Nobody can tell you that for your site, and this page makes no traffic claims. Record a baseline for your key pages in Search Console and compare over time. ### Can I see AI Mode traffic in Search Console? Not separately. It is counted in the Performance report under the Web search type, and a click on a link to an external page in AI Mode counts as a click. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Help: Get AI-powered responses with AI Mode in Google Search](https://support.google.com/websearch/answer/16011537?hl=en) - [Search Console Help: What are impressions, position, and clicks?](https://support.google.com/webmasters/answer/7042828?hl=en) - [Google Search Central: Robots meta tag, data-nosnippet and X-Robots-Tag specifications](https://developers.google.com/search/docs/crawling-indexing/robots-meta-tag) - [Google: List of Google's common crawlers (Google-Extended)](https://developers.google.com/search/docs/crawling-indexing/google-common-crawlers) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: How do you create an llms.txt file? A step-by-step guide, and what it can and cannot do Canonical: https://whatisgeo.app/how-to-create-an-llms-txt-file/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Write a plain Markdown file called llms.txt with an H1 title, a short summary and lists of links, and publish it at your site root. It is a proposed convention, and no platform guarantees it affects AI answers. The whole job takes an afternoon: pick a short list of your best pages, write them up in the format below, and get the file to load at yourdomain.com/llms.txt. Before you start, check whether your platform already makes one for you. Shopify does by default, and Yoast SEO can do it on WordPress. This page keeps three kinds of claim apart: what the llms.txt proposal itself says, what platforms document about the file, and what is only our recommendation. ## What the llms.txt format actually requires The format comes from the proposal at llmstxt.org, currently labelled v2. Its changes page dates v2 to August 2026 and says the original was published in September 2024. The text is maintained in Answer.AI's llms-txt repository on GitHub, and the site describes it as open for community input. According to the spec, a file in the format contains these parts, in this order: 1. An optional byte-order mark. 2. An H1 with the name of the project or site. **This is the only required section.** 3. A blockquote with a short summary, "containing key information necessary for understanding the rest of the file". 4. Zero or more sections of any Markdown except headings (paragraphs, lists) with more detail. 5. Zero or more sections under H2 headings, each holding a "file list" of links. Every item needs a Markdown link, `[name](url)`, and can then take a colon and a note. An H2 called "Optional" is, by convention, for secondary links an agent can skip when it needs a shorter context. v2 says the Optional section no longer has any special mechanical meaning. It is still a useful convention. Three other details from the spec are worth knowing: - **Location.** The file can sit at the root, `/llms.txt`, or at any path, such as `/docs/llms.txt`. A file covers the pages under its path, and where more than one applies, agents should use the most specific. - **What the links should point to.** The spec says links "should point to LLM-friendly content", such as clean Markdown versions of pages. Webflow's help centre takes a looser view: it says a well-structured HTML page linked from the file "should be sufficient for LLMs today". That is Webflow's opinion, and the spec does not say it. - **Inference rather than training.** The spec says llms.txt is used "on demand, when an agent needs information about a topic while assisting a user". It adds that the expectation was that it "would mainly be useful for inference rather than training", "though training runs could take advantage of the information too". ## Step 1: Decide which pages belong in it Start with a list rather than a file. Ask which pages you would hand to someone who needed to describe your business accurately in five minutes. What follows is our recommendation. No spec or platform requires it. Link to the canonical pages that best describe what you do: main product or service pages, pricing or ordering information, an about page, contact and location details, and any help content customers rely on. Leave out thin pages, duplicates, tag archives, internal or sensitive URLs, and anything you would not want quoted back to you. Keep it short. A dozen well-chosen links tell a clearer story than a sitemap pasted into Markdown. The spec makes a related point: the file "stays small enough to fit in context", and the detail lives behind the links. ## Step 2: Write the file Write it in a plain text editor and save it as `llms.txt` with UTF-8 encoding: 1. An H1 with the name of the business or site. 2. A blockquote with a one or two sentence summary in plain words. 3. Optional paragraphs, if a reader needs to know something before the links make sense. 4. H2 sections grouping your links by purpose. 5. Under each H2, a list of links, each followed by a colon and a short note. 6. If you need one, an "Optional" H2 for lower-priority links. The spec's writing guidelines are brief: use concise, clear language, give links short informative descriptions, and avoid ambiguous terms or unexplained jargon. Webflow's advice points the same way: say what your product actually does, and do not paste in homepage marketing copy. A good test for each note is whether it could sit on any company's site. If it could, rewrite it. ## A worked example Here is the file for an invented business, a small bakery with an online shop. Every name and URL is made up. ``` # Example Bakery > Independent bakery in Leeds selling sourdough, pastries and celebration cakes. Order online for collection in store or local delivery. Celebration cakes need at least five days' notice. We do not offer gluten-free products. ## Shop - [Bread and pastries](https://example.com/shop/bread): Full range with ingredients and allergen information - [Celebration cakes](https://example.com/shop/cakes): Sizes, flavours, prices and how to order ## About - [Our story](https://example.com/about): Who runs the bakery and how we bake - [Visit us](https://example.com/visit): Address, opening hours and collection details - [Delivery and refunds](https://example.com/delivery): Delivery area, charges and refund policy ## Optional - [Blog](https://example.com/blog): Recipes and bakery news ``` The file has no keyword lists, no "best bakery" claims and no forty-link dump. The paragraph after the summary uses the slot the spec sets aside for detail: it holds facts a reader needs before following the links. ## Step 3: Publish it at your site root Chrome's Lighthouse documentation says to create the file and "place it in the root directory of your website (for example, https://example.com/llms.txt)". That is the right target for a whole-site file. How you get it there depends on the platform. ### WordPress If you use Yoast SEO, its help centre documents a built-in llms.txt feature. In the WordPress dashboard, go to Yoast SEO, then Settings, then Site Features. Under AI tools, find LLMS.txt, switch the toggle on and click Save changes. Yoast says the file is generated and refreshed automatically. To choose the pages yourself, open Customize llms.txt file and select Manual page selection. Yoast notes the feature is not available for multisite, and that if you see "Your llms.txt file couldn't be auto-generated" you should check the web server's file system permissions. The llmstxt.org integrations list also names AIOSEO as a WordPress plugin with a generator. We have not checked AIOSEO's own documentation. Without a plugin, the generic route is to upload your `llms.txt` to your site's root folder on the server, using your host's file manager or SFTP, so it loads at /llms.txt. Our advice is to use one method only, so a plugin and a hand-made file are not both trying to serve the same URL. ### Shopify Shopify's developer documentation says that "by default, Shopify manages /llms.txt" by mirroring the content of /agents.md, and that "for most stores, the managed file is all you need". So open yourstore.com/llms.txt before you build anything. Your store probably already serves one. To change it, Shopify recommends an `agents.md.liquid` theme template, which customises /agents.md, /llms.txt and /llms-full.txt together. Add an `llms.txt.liquid` template only if you need /llms.txt to differ from /agents.md. In the admin, go to Online Store, then Themes, click the three dots on your theme and choose Edit code. Then right-click the Templates folder, choose New File and name it `llms.txt.liquid`. Shopify's documented limits: - It must be `llms.txt.liquid`. It cannot be a JSON template. - It accepts plain text or Markdown plus Liquid, but only the `request` and `agents` objects are available. Standard objects such as `shop` and `collections` are not. - Once you add it, you maintain its content yourself. - The file is served at the bare primary domain, with no locale or Markets subfolder. ### Webflow and other hosted builders Webflow's help centre documents an upload. Go to Site settings, then SEO, then LLMs.txt, click Upload file, choose your file and click Save changes. The file must be UTF-8 encoded and under 100 KB. After you publish, it is available at /llms.txt on your custom domain. It is not published to the Webflow staging domain. The llmstxt.org integrations list says Wix generates an llms.txt file for every Wix site. That claim comes from the spec site, and we have not checked it against Wix's own help centre. For any other builder, look in its official help for a way to serve a plain text file at the root path, then confirm the result in a browser. ## Step 4: Check that it works (a validation checklist) You do not need a validator tool. A browser and ten minutes are enough. This checklist is our recommendation, built from the spec's format rules. 1. Open yourdomain.com/llms.txt on the live domain. It should show plain text, not a styled web page or an error. 2. Check the first line is an H1 (`# Name`), followed by the `>` summary if you wrote one. 3. Check that the only headings after that are H2s (`##`), and that each list item starts with a `[name](url)` link. 4. Open every link. Each should load the page you meant and should not return an error. 5. Check that robots.txt does not block those pages for the crawlers you care about. Our [crawler access](/guide/#crawler-access) section covers the main user agents. 6. Reread the notes. Do they still describe each page accurately? 7. Run the spec's own test: give an AI assistant only your llms.txt and ask it questions about your business. Lighthouse has an llms.txt audit, but it checks little. Its documentation says it flags the page if a server error occurs when fetching the file, and marks the audit not applicable if the file returns a 404, "as providing the file is optional at the moment". It describes no check of the file's format, so passing the audit does not mean the file is well formed. ## Should you use a generator? A generator or plugin saves typing, and on Shopify, or with Yoast switched on, one may already be running. The trouble is that automatic tools tend to include whatever they find, while the point of the file is a short, curated list. Our recommendation: treat generated output as a first draft, cut it down, rewrite vague notes and run the checklist above. We have not tested any standalone generator, so we do not name or rank any. ## What llms.txt does not do, and what is documented about it No one can guarantee that this file changes whether or how an AI assistant cites you. Here is what the sources we checked actually say. - **The spec** presents llms.txt as a proposal for helping agents while they assist users. It says thousands of sites publish one and that the files are used most heavily for software documentation. Those are the spec's own statements. - **Google's** AI features documentation says that to be a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to show with a snippet, and "there are no additional technical requirements". It also says "specific optimization isn't required". It does not mention llms.txt. - **OpenAI, Anthropic and Perplexity** describe their bots and robots.txt controls in their crawler documentation. None of those pages says the crawlers read a website's llms.txt. OpenAI and Perplexity publish llms.txt files for their own developer docs. That shows the format in use for documentation. It does not show that ChatGPT or Perplexity read your file when answering. - **Webflow** calls the format "experimental", says it "doesn't guarantee your site will be read by LLMs", and says its effects on AI-driven search "are unclear". So treat llms.txt as cheap and low-risk, and do not mistake it for a ranking lever. Whether assistants describe you well depends on the wider work covered in [what GEO is](/guide/#what-is-geo) and [what AI visibility means](/what-is-ai-visibility/). To find out whether anything you change makes a difference, see our section on [measurement](/guide/#measurement). ## Keeping the file up to date An out-of-date llms.txt points at pages that have moved and describes a business that has changed. Webflow's advice is to update it regularly as your site changes, and we agree. In practice, review it whenever you add or retire a product line, change prices or policies, or restructure a section, and run the checklist again each time. If Shopify or a plugin generates the file for you, still read what it actually serves, because that is what an agent will see. ## From the same studio These are made by the studio that publishes this guide. - [Hilyt](https://hilyt.it/): If you want a structured, AI-readable description of your business alongside your llms.txt, Hilyt publishes an open profile for it. - [visibilitytoai](https://visibilitytoai.com/): llms.txt carries no guarantee, so the useful next step is checking whether ChatGPT, Claude, Gemini and Perplexity actually name and cite you. ## Questions people ask ### How do I create an llms.txt file? Write a Markdown file with an H1 title, an optional blockquote summary and H2 sections listing links to your key pages, then publish it so it loads at yourdomain.com/llms.txt. Check first whether your platform already generates one, as Shopify does by default. ### What is the llms.txt format? According to llmstxt.org, it is Markdown in a fixed order: an H1 with the site name (the only required part), a blockquote summary, optional detail without headings, then H2 sections of links written as `[name](url)` with an optional colon and note. An H2 called Optional is a convention for secondary links. ### Are llms.txt files worth it? The file is cheap to make and low risk, but none of the crawler documentation we checked from Google, OpenAI, Anthropic or Perplexity says it is used for answers, and Webflow calls its effects on AI-driven search unclear. Treat it as a small extra, not a strategy. ### Is llms.txt still used? Yes, as a proposal. The spec reached v2 in August 2026, Lighthouse audits for it, and Shopify, Webflow and Yoast all support it. None of the major AI platforms' crawler documentation commits to reading it. ### How do I add llms.txt to WordPress or Shopify? On WordPress, Yoast SEO can generate one under Settings, Site Features, AI tools, LLMS.txt, or you can upload a file to your site root. On Shopify, the platform serves /llms.txt by default, and you customise it through an agents.md.liquid or llms.txt.liquid theme template. ## Sources - [llmstxt.org: The /llms.txt file, v2](https://llmstxt.org/) - [llmstxt.org: Changes (v2, August 2026)](https://llmstxt.org/changes.html) - [Chrome for Developers: Lighthouse llms.txt audit](https://developer.chrome.com/docs/lighthouse/agentic-browsing/llms-txt) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Shopify developer docs: llms.txt.liquid](https://shopify.dev/docs/storefronts/themes/architecture/templates/llms-txt-liquid) - [Webflow Help Center: Upload an llms.txt file to your site](https://help.webflow.com/hc/en-us/articles/43240104183315-Upload-an-llms-txt-file-to-your-site) - [Yoast: How to enable llms.txt with Yoast SEO](https://yoast.com/help/enable-llmstxt/) - [Yoast: llms.txt feature page](https://yoast.com/features/llms-txt/) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Anthropic: Does Anthropic crawl data from the web?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) --- # Guide: How do you measure AI visibility? A method you can run and trust Canonical: https://whatisgeo.app/how-to-measure-ai-visibility/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Ask each assistant the same set of real buyer questions, repeat every run, and score whether you are named, cited and described correctly. Then cross-check with referral traffic, server logs and platform reports. No single check tells you where you stand, because the same question can get a different answer on the next run. So this method rests on repeated samples, scored the same way every time and checked against data you already own. You can do all of it by hand with a spreadsheet. The five steps below take you from a blank sheet to a baseline you can compare month to month. Throughout, we label three kinds of claim: what a platform documents, what research has found under its own conditions, and what is our recommendation. The recommendations are sensible practice. None of them is proof. ## What you can and cannot measure You can measure: - whether your business is **named** in an answer to a given question, on a given day, under given conditions - whether the answer **cites** a page on your site - whether what it says about you is **accurate** - tagged and referred visits from assistants in your analytics - crawler and user-triggered visits from AI companies in your server logs - the figures Google and Bing report to site owners You cannot measure a stable position. Google says AI Mode and AI Overviews "may use different models and techniques, so the set of responses and links they show will vary", and that AI Overviews often don't appear at all. Bing says its own AI Performance report "does not measure rankings, authority, performance, or importance". Treat every figure as a sample from a system that keeps moving. For how these answers get put together in the first place, see [how AI answers happen](/guide/#how-it-works). ## Step 1: Build a fixed prompt set from real buyer questions Start with questions your customers actually ask. Pull them from sales calls, support emails, contact form messages, reviews and the search terms you already see. Write them the way a buyer would type them, messy phrasing included. A prompt that reads like a marketing brief won't behave like a real question. **Our recommendation:** aim for 15 to 30 prompts. That is a suggestion, not a research finding. It is enough to cover your main services and buying stages, and few enough that a small team will actually finish each round. The [30-day plan](/guide/#getting-started) on our home guide starts with 3 to 5 prompts, which is a good first look. Widen the set once the routine sticks. Mix three types: - **Unbranded:** "Who does emergency boiler repair in Leeds?" These matter most, because they show whether you appear when nobody asks for you. - **Comparison:** "What should I look for when choosing an accountant for a small charity?" - **Branded:** "What does [your business] do?" These are your accuracy checks. Then freeze the list and give it a version number. If you change a prompt's wording, you have started a new test, and its results won't compare with the old ones. ## Step 2: Fix the conditions and run each prompt more than once Conditions change answers, so record them. **Our recommendation** is to log, for every run: - the date - the assistant, plus the mode or model if the app shows one - your location (or the location you set) - whether you were logged in Where you can, run each prompt in a fresh or temporary chat so that earlier questions don't shape later answers. Stick with the same assistants every round. Choose the ones your buyers plausibly use, and don't add a new one halfway through and compare it as though nothing changed. Then repeat. **Our suggestion** is at least three runs per prompt on each assistant to begin with, and more where answers swing about. There is no documented right number. For comparison, the researchers behind the GEO paper (below) ran their experiments on five different random seeds and averaged the results "to reduce variance". That was a lab setup with its own goals, so it is no target for you. It does show that people who study these systems don't trust a single run either. ## Step 3: Score each answer: named, cited, described correctly For every answer, record yes or no for three questions: 1. **Named.** Does the answer mention your business by name? 2. **Cited.** Does it link to a page on your site? If so, note which page. 3. **Described correctly.** Is everything it says about you accurate? Also note which competitors were named. That shows you who the assistant treats as the alternatives. Keep the scoring blunt. A ten-point scale invites you to argue with yourself, and you will score the same answer differently on another day. Track naming and citing separately: you can be named with no link, or linked while the text barely mentions you. Accuracy is the check people skip. OpenAI's own help page warns that ChatGPT search results and citations "can be incomplete, outdated, or incorrect". Being named with the wrong address, an old price or a service you no longer offer can do more harm than being left out. Copy the wrong statement word for word, so you can find and fix whatever it came from. **What the research says.** Aggarwal et al. (GEO, KDD 2024) proposed two visibility measures for generative engines. *Position-adjusted word count* counts the words in sentences that cite a source, weighted down the later the citation appears. *Subjective impression* has an LLM (GPT-3.5, using the G-Eval method) score seven aspects, such as how relevant the cited material is and how likely a user is to click it. The authors tested these on their own benchmark, GEO-bench (10,000 queries), with an engine they built that answered from the top five Google results, and ran a further check on Perplexity. These are research measures under the authors' conditions. No assistant reports them to you. The yes/no scoring above is a simpler version of the same idea that a small business can run. ## Step 4: Add the data you already own: referrals, server logs and platform reports Your own data is the cross-check. It doesn't depend on which prompts you wrote. ### What the platforms document **Google Search Console.** Google says sites appearing in AI Overviews and AI Mode are "included in the overall search traffic in Search Console", reported in the Performance report "within the 'Web' search type". Its AI features page describes no separate AI-only view in that report, so treat these as part of your overall search totals. Google also says there are "no additional technical requirements" for appearing in AI features: a page must be indexed and eligible to show in Search with a snippet. Your normal indexing checks still count here. **Bing Webmaster Tools.** Bing's AI Performance report, released as a public preview in February 2026, shows citation activity across Microsoft Copilot, AI-generated summaries in Bing and "select partner AI integrations". It reports total citations, average cited pages per day, page-level citation counts and grounding queries (the phrases the AI used when it retrieved content it then cited). Bing says the grounding query data "represents a sample of overall citation activity". **ChatGPT referrals.** ChatGPT links have carried a `utm_source=chatgpt.com` tag, so look for it in your campaign and referral reports. OpenAI's publisher FAQ was not available at its published address when we checked, so we are relying on industry reporting here. In October 2026 the tag and the referrer went missing from ChatGPT links for a few days, and those visits showed up as direct traffic. The Adobe analyst who spotted it later said it was a bug and that tagging had resumed (reported by Search Engine Roundtable). **Server logs.** The AI companies publish what each of their user agents does: | Company | User agent | Documented role | |---|---|---| | OpenAI | OAI-SearchBot | Surfaces sites in ChatGPT's search features | | OpenAI | GPTBot | Crawls content that may be used to train models | | OpenAI | ChatGPT-User | Visits a page when a user's question needs it; not an automatic crawler | | Perplexity | PerplexityBot | Surfaces and links sites in Perplexity search results | | Perplexity | Perplexity-User | Fetches a page when a user asks a question | | Anthropic | ClaudeBot | Collects content that could contribute to model training | | Anthropic | Claude-SearchBot | Indexes content to improve search results | | Anthropic | Claude-User | Fetches pages when people ask Claude questions | OpenAI, Perplexity and Anthropic each publish the IP addresses their bots use. Check against those lists before you trust a user-agent string, because anyone can copy one. If these bots can't reach your pages at all, fix that first: see [crawler access](/guide/#crawler-access). ### What to treat with caution - **A crawl is not a citation.** A search or training crawler fetching a page doesn't mean the page was used in any answer. The user-triggered agents (ChatGPT-User, Perplexity-User, Claude-User) are a closer signal, since they fetch because someone asked a question. Even so, a fetch doesn't prove you were cited. - **Tagged traffic is a minimum.** Visits can arrive with no tag and no referrer, as the October 2026 episode showed, and those get filed as direct. - **Trends don't explain themselves.** Bing says changes in its citation trends "are observational" and "cannot be attributed to a specific cause or event". The same caution applies to your own numbers. - **Rising together isn't cause.** If referrals and your mention rate rise in the same month, that is worth noting. It doesn't prove one caused the other. ## Step 5: Record, compare over time and read the changes honestly Keep one sheet with one row per run: | Column | Example entry | |---|---| | Prompt set version | v1 | | Date | the day you ran it | | Assistant and mode | ChatGPT, search on | | Location / logged in | Manchester / logged out | | Prompt | the exact wording | | Named / Cited / Accurate | Y / N / Y | | Page cited, competitors named, errors | free text | Summarise each prompt and assistant as a rate: named in *x* of *y* runs, cited in *x* of *y*, accurate in *x* of *y*. **Our recommendation** is to report rates or ranges and never a single result, and to run a round at a steady interval, monthly for most small businesses. Only compare like with like: the same prompt set version, the same assistants, the same conditions. If any of those change, note it in the sheet and don't draw a trend line across the break. Small moves between rounds are often noise. Before you call something a trend, look for a change that holds across several rounds and several prompts, and that your referral and log data agree with. More on reading progress is in [measuring progress](/guide/#measurement). ## Doing it by hand or with a tool Everything above works by hand: a frozen list of prompts, a spreadsheet and a couple of hours each round. Nothing here needs a paid tool. A tool can save time by running prompts on a schedule, across more assistants and locations, and logging the results. That is most useful once you know what you are measuring. Run at least one round by hand first. You'll learn what answers in your sector look like, which prompts matter and what counts as accurate. Then you can judge whether a tool's numbers mean anything, and ask how it handles the run-to-run variation described above. ## What a good result looks like (and what no measurement can promise) A good result is modest and steady. You are named in a fair share of unbranded answers, citations point to the right pages, descriptions are accurate, and that holds across rounds as well as in one lucky run. No measurement can promise inclusion, and nobody can guarantee that an assistant will name you. What you can do is make your site easy to crawl, easy to understand and accurate, then use this method to see whether things move. The [core principles](/guide/#principles) cover the first part. The rest is careful, repeated sampling. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you know what you want to track, it runs buyer questions across ChatGPT, Claude, Gemini and Perplexity and records whether your company is named and cited, so you don't have to rerun them by hand. ## Questions people ask ### How do I check if ChatGPT mentions my business? Ask it the questions your buyers ask, mostly without naming your business, in a fresh or temporary chat. Run each question more than once and record whether you are named, cited and described correctly. A rate across runs tells you far more than one answer. ### How many times should I run each prompt to trust the result? No platform documents a number. Our suggestion is at least three runs per prompt on each assistant to start, with more where the answers change a lot. If the same brands appear every time you can be more confident; if they move around, report a rate and say so. ### Can I see AI Overviews traffic in Google Search Console? Yes, as part of the total. Google says sites that appear in AI Overviews and AI Mode are counted in the overall search traffic in the Performance report, under the Web search type. Its AI features page does not describe a separate AI-only view in that report. ### How do I find ChatGPT or Perplexity referral traffic in analytics? Check your referral sources for the assistants' domains and, for ChatGPT, campaign data tagged utm_source=chatgpt.com. That tag went missing for a few days in October 2026 before it was reported fixed, so read tagged traffic as a minimum figure. ### Is there a single AI visibility score I can rely on? No. Answers vary between runs and between features, and Bing says its own AI Performance report does not measure rankings. Track mention, citation and accuracy rates over repeated runs, and look at how they change over time. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Bing Webmaster Tools help: AI Performance](https://www.bing.com/webmasters/help/ai-performance-9f8e7d6c) - [Bing Webmaster Blog: Introducing AI Performance in Bing Webmaster Tools Public Preview (February 2026)](https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [OpenAI Help Center: Searching the web with ChatGPT](https://help.openai.com/en/articles/9237897-chatgpt-search) - [Search Engine Roundtable: OpenAI drops UTM parameters from ChatGPT results (14 October 2026, updated)](https://www.seroundtable.com/utm-parameters-missing-chatgpt-42299.html) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic (Claude Help Center): Does Anthropic crawl data from the web?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024 (arXiv)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: How do you rank in ChatGPT? What is documented, what is research, and what is guesswork Canonical: https://whatisgeo.app/how-to-rank-in-chatgpt/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > ChatGPT has no ranking to climb. When it searches the web it picks and cites sources; you can improve your odds by letting OAI-SearchBot in and publishing clear, findable pages, but nobody can guarantee a mention. "Rank" is a search engine word. Google shows an ordered list of results. ChatGPT gives one answer, sometimes with citations, and OpenAI's own help centre says that placement in its search results is not guaranteed. A better question is what makes your site eligible to be cited, and what improves the odds once it is. Most of the top results on this topic are agency and vendor listicles: confident tips with little sourcing. This page puts each tip in one of three groups. **Documented** means OpenAI publishes it. **Researched** means a peer-reviewed paper tested it, and we say under what conditions. **Recommended** is our own advice, labelled as such. ## Two ways ChatGPT can mention you: training and search ChatGPT can answer from what its model learned in training, or it can search the web first. OpenAI's help centre says ChatGPT may search automatically when a question would benefit from current information, and users can also choose Search themselves. Responses that use web search may include citations, and a **Sources** view lists the cited pages and other relevant links. This split matters because only the search route gives your page a citation someone can click. Content can also reach a model through training, which OpenAI's GPTBot crawler collects for. That route is slow and you cannot see it: you have no way to check what a model learned from you, or when. If you want a named, linked mention, prepare for the search route. For the wider picture of how AI answers are put together, see [how AI answers happen](/guide/#how-it-works). ## What OpenAI documents OpenAI's crawler page lists the user agents it uses and what each one is for: | User agent | What OpenAI says it does | What robots.txt does | | --- | --- | --- | | OAI-SearchBot | Surfaces websites in ChatGPT's search features | Sites that opt out "will not be shown in ChatGPT search answers", though they can still appear as navigational links | | GPTBot | Crawls content that may be used to train OpenAI's foundation models | Disallowing it signals that your content should not be used for training | | ChatGPT-User | Visits a page when a user's request needs it; it does not crawl automatically | OpenAI says robots.txt rules "may not apply", and this agent is not used to decide what appears in search | | OAI-AdsBot | Checks landing pages submitted as ads on ChatGPT | Only visits pages submitted as ads | Three more documented points matter here: - **Each setting is independent.** OpenAI's example is a site that allows OAI-SearchBot so it can appear in search results, while disallowing GPTBot so its content is not used for training. - **Changes are not instant.** OpenAI says it can take around 24 hours after a robots.txt update for its search systems to adjust. - **Eligibility has two parts.** The help centre says that to make a site eligible, you should allow OAI-SearchBot to crawl it and confirm that your host or content delivery network (CDN) allows traffic from OpenAI's published search bot IP addresses. On ranking itself, OpenAI says only that ChatGPT "ranks search results using multiple factors intended to help users find relevant, reliable information" and that placement is not guaranteed. It does not publish those factors. Anyone selling a ChatGPT "ranking formula" is guessing, whatever they call it. ## Step 1: make sure OAI-SearchBot can reach you This is the one step OpenAI documents directly, so do it first. 1. **Read your robots.txt.** It lives at the root of your domain (yourdomain.com/robots.txt). Look for a group naming `OAI-SearchBot`, and for a catch-all `User-agent: *` group with `Disallow: /`. Templates that block every AI crawler at once often catch OAI-SearchBot along with the training bots. 2. **Decide on each bot separately.** A site that wants to appear in ChatGPT search but keep its content out of training might use: ``` User-agent: OAI-SearchBot Allow: / User-agent: GPTBot Disallow: / ``` That choice is yours. This example shows the two settings working independently. It is not a recommendation to block training. 3. **Check your host or CDN.** Bot protection or firewall rules can turn away a crawler that robots.txt allows. OpenAI publishes the search bot's IP ranges at openai.com/searchbot.json and asks that they be allowed. 4. **Allow around a day** for OpenAI's search systems to pick up a change, as its documentation says. One caveat about robots.txt in general. RFC 9309, the standard behind it, says its rules "are not a form of access authorization". Listing a path in robots.txt also makes that path public. If something must stay private, protect it with real access control, such as a login. Our guide to [AI crawler access](/guide/#crawler-access) covers the other platforms' bots. ## Step 2: be easy to find and read in ordinary search OpenAI's help centre says ChatGPT search sometimes works with other search providers. It usually rewrites the user's question into one or more targeted queries and sends those to the provider. The article lists Microsoft's privacy statement among the providers' policies. Our reading, which OpenAI does not state in so many words, is that pages ordinary search engines can find and understand give this process more to retrieve. That makes the groundwork familiar: pages that can be crawled and indexed, a clear title and topic for each page, and important text in the HTML itself rather than loaded later by scripts. Where this overlaps with traditional SEO and where it differs is covered in [GEO vs SEO](/guide/#geo-vs-seo). ## What research suggests about content that gets cited The study most often quoted here is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024. Its conditions shape what you can take from it. **What they tested.** Nine ways of rewriting a source's text: making it more authoritative in tone, adding statistics, keyword stuffing, citing sources, adding quotations, simplifying the language, improving fluency, adding unique words and adding technical terms. **How they tested it.** The authors built their own generative engine. For each query it fetched the top five Google results, and a language model wrote an answer from them. The queries came from GEO-bench, their benchmark of 10,000 queries drawn from nine sources across 25 domains. They measured "visibility" with their own metrics: how much of the answer drew on a source and how prominently, plus a subjective impression score judged by GPT-3.5. They also repeated the methods on Perplexity.ai. **What they found.** In their setup, citing sources, adding quotations and adding statistics did best. The authors' headline is a visibility boost of up to 40% on their metric. Improving fluency and readability also helped. Keyword stuffing, a classic SEO tactic, gave little or no improvement. The authors also say results varied by domain. **What it does not show.** - It did not test ChatGPT. Do not treat "works in ChatGPT" as a finding of this paper. - It measured how visible a source was among five pages already retrieved. It does not tell you how to get retrieved in the first place, which is the job of Steps 1 and 2. - "Visibility" is the authors' own measure, partly scored by another language model. It does not measure traffic or sales. - Generative engines change often, so a 2024 benchmark result is evidence, not a lasting rule. The reasonable takeaway is modest. How a page is written can affect whether a generative engine uses it. Adding real, sourced statistics and quotations is worth considering. Making numbers up to game this would be dishonest, and it would undermine trust in your page. ## Sensible practice, labelled as recommendation OpenAI does not document the points below, and the paper did not test them on ChatGPT. They are our advice. 1. **Write pages that each answer one question.** A page with one clear job is easier for a reader to use and easier for a search system to match to a query. 2. **Keep key facts in plain text.** Prices, services, locations and opening hours should sit in the page's HTML, not only in images or content loaded by scripts. 3. **Keep facts consistent.** Your site and your profiles elsewhere should describe what you do in the same way. Contradictions confuse people first and machines second. 4. **Test, then record what you find.** See the next section. None of this is clever, and it helps human readers whether or not a model ever cites the page. The reasoning behind these points is set out in our [GEO principles](/guide/#principles). ## What you cannot control, and claims to ignore You cannot control whether ChatGPT decides to search for a given question, which sources it picks or how it weighs them. Answers also vary by user. OpenAI says ChatGPT may use a person's approximate location and, if memory is turned on, their saved memories when rewriting a search query. Treat these claims as red flags: 1. **A guaranteed mention or position.** OpenAI itself says placement is not guaranteed. 2. **Precise percentages or timelines with no named study.** If you are not told who measured it and how, the number tells you nothing. 3. **"Proven in ChatGPT" claims based on research done elsewhere.** The GEO paper is useful, but it is not a ChatGPT study. 4. **"Official" tips you cannot find on OpenAI's own pages.** 5. **Paid placement in the answer itself.** OpenAI documents ads on ChatGPT, but nothing we read says that buying ads changes which sources ChatGPT search cites. If someone quotes a vendor's study, attribute it to that vendor and ask what it is selling. ## How to check where you stand, by hand You do not need a tool for a first look. 1. **List your buyers' real questions**, the ones that come in by phone, email and sales calls, in their words. 2. **Ask each one in a fresh conversation.** OpenAI says people who are not signed in can also use web search. That lets you check without your own memory shaping the query. Location still applies, so note where you ran the check. 3. **Record the same fields each time:** date, question, whether ChatGPT searched (a Sources list appears), whether it named your business, whether it cited your site, and which competitors or sites it named instead. 4. **Open the cited sources.** OpenAI's help centre warns that citations can be incomplete, outdated or wrong, and you may find ChatGPT describing you from someone else's out-of-date page. 5. **Repeat with the same list later.** Answers vary between runs, so one check proves little. A pattern across many questions tells you more. To turn this into a regular routine, see [measuring progress](/guide/#measurement). ## Where to go next Start with Step 1, because it is quick and it is the one thing OpenAI documents as a requirement. Then run the hand check, so you know where you stand before you change any content. Changes to robots.txt, your CDN rules or your pages can then be judged against that record, not against a vendor's promise. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you have checked a few questions by hand, it tracks whether ChatGPT and other assistants name and cite your company for your buyers' questions over time. ## Questions people ask ### How do I get my website to show up in ChatGPT answers? You can make your site eligible and improve your odds, but you cannot force it. OpenAI says to allow OAI-SearchBot in robots.txt and let its published IP addresses through your host or CDN. After that, clear pages that ordinary search can find give ChatGPT search something to cite. ### Does blocking GPTBot stop me appearing in ChatGPT search? Not according to OpenAI. Its crawler documentation says each robots.txt setting is independent, and gives the example of allowing OAI-SearchBot to appear in search results while disallowing GPTBot so content is not used for training. ### How can I tell if ChatGPT is mentioning my business? Ask it your buyers' real questions and record whether it searched, whether it named you, whether it cited your site and who it named instead. Repeat the same questions over time, because a single answer tells you little. ### Is SEO for ChatGPT different from normal SEO? It overlaps a lot. OpenAI says ChatGPT search sometimes uses other search providers, so being findable in ordinary search still matters, and there is one extra crawler to allow: OAI-SearchBot. The research on writing for AI answers is early and was not done on ChatGPT. ### Can I pay to rank higher in ChatGPT? OpenAI's documentation mentions ads on ChatGPT, but nothing we read says paying changes which sources ChatGPT search cites. OpenAI's help centre says placement in search results is not guaranteed, so be wary of anyone who sells guaranteed placement. ## Sources - [OpenAI: Overview of OpenAI Crawlers](https://platform.openai.com/docs/bots) - [OpenAI Help Center: Searching the web with ChatGPT](https://help.openai.com/en/articles/9237897-chatgpt-search) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) --- # Guide: How do you rank in Google AI Overviews? Canonical: https://whatisgeo.app/how-to-rank-in-google-ai-overviews/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Google says there is no special optimisation for AI Overviews. A page must be indexed and eligible for a snippet, and the usual helpful-content and technical SEO basics apply. Nobody can guarantee inclusion. There is no position inside an AI Overview to climb, and no tag, form or setting that asks Google to include you. What you can control is whether a page qualifies to be used as a supporting link, and Google spells out those conditions in its own documentation. Most popular guides on this topic present guesses and correlation studies as proven tactics. This page keeps three kinds of claim apart and labels each one: what Google documents, what research suggests, and what we recommend. For background on how AI answers are put together in general, start with [how AI answers happen](/guide/#how-it-works). ## What Google itself documents Google's Search Central page "AI features and your website" covers AI Overviews and AI Mode from a site owner's point of view. These are its key statements: - **Eligibility.** To be eligible as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet, meeting the normal Search technical requirements. Google adds: "There are no additional technical requirements." - **No special optimisation.** Google says specific optimisation isn't required for AI Overviews and AI Mode, and that existing SEO fundamentals continue to be worthwhile. - **AI Overviews often don't appear.** Google says they are only shown when its systems decide they add something beyond classic Search, "and as such, often don't trigger." - **The two features differ.** AI Mode and AI Overviews may use different models and techniques, so the responses and links they show will vary. The third point matters. For many queries there is no AI Overview at all, so there is nothing to appear in. ## Eligibility checklist: indexed, snippet-eligible, crawlable Google's wording turns into a short checklist. Each item comes from the AI features page or Google's May 2025 Search Central blog post on succeeding in AI search. | Condition | What to check | | --- | --- | | Indexed | The page is in Google's index. Verify your site in Search Console, which Google recommends for diagnosing technical issues, and look for a forgotten `noindex`. | | Snippet-eligible | Nothing on the page stops Google showing a snippet, such as a `nosnippet` rule. | | Crawlable | robots.txt allows crawling, and your CDN or hosting is not blocking Googlebot. | | Working page | Google receives an HTTP 200 status and the page has indexable content. | | Text available | Important content is available in textual form. | None of this is glamorous, but it is the only part with documented rules. For which crawlers to allow and why, see [crawler access](/guide/#crawler-access). ## What Google says about content and page experience Google's May 2025 post, by John Mueller of Google Search Relations, and the AI features page give the same list of fundamentals: - Unique, non-commodity content that people will find helpful and satisfying. - A good page experience: it displays well across devices, responds quickly, and makes the main content easy to tell apart from everything else. - Content that is easy to find through internal links on your site. - High-quality images and videos supporting the text, where they fit. - Structured data that follows Google's guidelines and matches the visible text on the page. - Up-to-date Merchant Center and Business Profile information, which matters mainly if you sell products or serve a local area. Notice what is missing. Google does not ask for a particular word count, a question-and-answer layout, a special file or a new type of markup. Some of those may still help readers, but Google has not documented any of them as a requirement for AI features, so treat claims that they are required as unverified. The overlap with ordinary SEO is covered in [GEO vs SEO](/guide/#geo-vs-seo). ## How AI Overviews choose sources: query fan-out Google says AI Overviews and AI Mode may use a "query fan-out" technique, "issuing multiple related searches across subtopics and data sources" to develop a response. While the response is generated, its models identify more supporting pages, which Google says lets it display "a wider and more diverse set of helpful links" than a classic web search. In practice, that means: - The links in an AI Overview can differ from the classic results for the same query. - A page that answers one related subtopic well may be relevant even if it is not near the top for the exact phrase typed. - Checking only where you rank for one keyword gives you an incomplete picture. Keep to Google's wording here. It says "may use", and it does not publish how often fan-out happens or which queries trigger it. Nobody outside Google can tell you that with authority. ## What research and third-party studies suggest, and their limits The best-known paper is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024. The authors built a benchmark of queries called GEO-bench and tested content changes such as adding citations, quotations from relevant sources and statistics. They report visibility gains of up to 40% in generative engine responses, gains of up to 37% in a test on Perplexity.ai, and that how well each method worked varied by domain. The limits matter as much as the findings: - The main experiments used the authors' own generative engine: an LLM writing answers from the top five Google search results for each query. That is a research setup, not Google's AI Overviews. - Visibility was measured with metrics the paper defines, which are not the same as being shown as a link in Google. - It is not a study of Google AI Overviews, so applying its numbers to Google is an assumption. Agencies and SEO software vendors also publish studies of which pages AI Overviews cite. We have not relied on any of them here. If you read one, look for who ran it, how many queries it covered, when, in which country, and whether it shows a correlation or a tested cause. A headline percentage from a vendor is that vendor's finding, not a Google rule. ## Sensible practice: our recommendations This section is our opinion, built on Google's guidance above. It is not a Google requirement and it guarantees nothing. 1. **Fix eligibility first.** Check your important pages in Search Console, look for stray `noindex` or `nosnippet` rules, and confirm your CDN is not blocking Googlebot. It is cheap and it is the only part with documented rules. 2. **Answer the question plainly, in text.** Put a direct answer near the top of the page, then the detail. This serves readers, which is what Google's guidance points to. 3. **Cover the follow-up questions.** Because of fan-out, a page that handles the related subtopics a reader would ask next has more chances to be relevant. 4. **Keep structured data honest.** If you use it, make sure it matches what is visible and validate it. 5. **Keep business listings current.** If you sell products or serve a local area, keep Merchant Center and Business Profile up to date. The thinking behind these is set out in [our GEO principles](/guide/#principles). ## Controls: nosnippet, data-nosnippet, max-snippet and noindex Google says AI is built into Search, so the controls for AI features are the normal Search controls. Its documentation names these: | Control | What it does | | --- | --- | | `nosnippet` | Stops Google showing a text snippet for the page. | | `data-nosnippet` | An HTML attribute that keeps specific parts of a page out of snippets. | | `max-snippet:[number]` | Sets a maximum snippet length. | | `noindex` | Keeps the page out of Google's index, and so out of Search entirely. | Google says "more restrictive permissions will limit how your content is featured in our AI experiences." These controls apply to Search as a whole, so `nosnippet` also removes your snippet from classic results. Make that trade deliberately. Two further points from Google's documentation: - **Google-Extended is a different control.** Google describes it as a robots.txt token for managing whether content may be used to train future Gemini models and for grounding in Gemini Apps and Vertex AI. Google states it "does not impact a site's inclusion in Google Search", and its AI features page names Googlebot's robots.txt rules as the crawling control for Search. Blocking Google-Extended does not take you out of AI Overviews. - **Changes take time.** If your content still appears after you add a preview control, Google suggests using the URL Inspection tool to check that Googlebot received the control, then allowing time to recrawl, which it says can take anywhere from several days to several months. You can also request a recrawl. ## How to check whether you appear and measure it - **Spot checks.** Search the questions your customers ask and look. AI Overviews often don't trigger and the links shown vary, so treat this as a spot check, not a measurement. - **Search Console.** Google says sites appearing in AI features are included in overall search traffic in the Performance report, under the "Web" search type. Its AI features page does not describe a separate AI Overviews report, so expect AI clicks and impressions to sit inside your existing totals. - **Your own analytics.** Google suggests tracking conversions and time on site as well as clicks. It also says clicks from results pages with AI Overviews are "higher quality", meaning people are more likely to spend more time on the site. That is Google's own observation, and the page does not publish data behind it. Be wary of any tool that claims to split AI Overview traffic out of Search Console totals without explaining how. More on tracking in [measuring progress](/guide/#measurement). ## What not to believe - **"Follow these steps to rank in AI Overviews."** Google documents eligibility, not a ranking you can climb. - **"We guarantee you a place."** Google does not promise inclusion, so nobody else can. - **"This tactic is proven."** If the proof is a correlation study or a test on a different engine, it is not proof for Google. - **"Block Google-Extended to stay out of AI Overviews."** Google says Google-Extended does not affect inclusion in Search. - **"You need a special file or markup for AI Overviews."** Google says there are no additional technical requirements. A claim is only as good as its source, so ask where it comes from. ## Questions people ask ### Is there a way to rank in Google AI Overviews? Not in the sense of climbing to a position. Google says a page must be indexed and eligible to be shown with a snippet, and that there are no additional technical requirements. Beyond that, nobody can guarantee inclusion. ### Do I need to do anything different from normal SEO to appear in AI Overviews? Google says specific optimisation isn't required for AI Overviews and AI Mode. Its guidance points to the same fundamentals as before: helpful, people-first content, good page experience, crawl access and structured data that matches the visible page. ### Do I need to rank on page one to be cited in an AI Overview? Google does not document that requirement. It says AI Overviews may use query fan-out, running related searches across subtopics, and can show a wider set of links than a classic search, so supporting links can differ from the top classic results. ### Can I stop my content appearing in AI Overviews? Yes. Google says nosnippet, data-nosnippet, max-snippet and noindex control what is shown from your pages in Search, including AI features, but they also affect classic results. Blocking Google-Extended does not do it, because Google says that token does not affect inclusion in Search. ### Do schema markup or llms.txt help me appear in AI Overviews? Google says there are no additional technical requirements for AI features. It asks that any structured data matches the visible text on the page, and its AI features documentation does not mention llms.txt. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central Blog: Top ways to ensure your content performs well in Google's AI experiences on Search (May 2025)](https://developers.google.com/search/blog/2025/05/succeeding-in-ai-search) - [Google Search Central: Control your snippets in search results](https://developers.google.com/search/docs/appearance/snippet) - [Google: List of Google's common crawlers (Google-Extended)](https://developers.google.com/search/docs/crawling-indexing/google-common-crawlers) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: How do you show up in Claude? What is documented, what is research, and what is guesswork Canonical: https://whatisgeo.app/how-to-show-up-in-claude/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Claude names businesses from its training data or, when it searches the web, from live pages it cites. Anthropic documents its crawlers and search, but nobody can guarantee a mention. Claude can mention you through two routes: what a model learned in training, and what it finds when it searches the live web. Anthropic documents the crawlers behind both routes and how to control them with robots.txt. It publishes nothing about how Claude chooses which businesses to name. So this page labels every claim. **Documented** means Anthropic or a standards body publishes it. **Researched** means a peer-reviewed paper tested it, under conditions we state. **Recommended** is our own advice. ## Two ways Claude can mention you: training data and web search **Training data.** Anthropic's help centre says each Claude model has a knowledge cutoff, and that models "may not be aware of events or information that occurred after their respective cutoff dates." So if you rename a service or change your prices today, Claude's built-in knowledge will not reflect it. That only changes when a model trained on later data comes out, and you have no way to request that. **Web search.** Claude's help centre says that when you ask about topics that benefit from current information, Claude "invokes a search tool" to ground its answer in content from the live web, and that every response includes citations. It describes those as direct citations to sources, source links for further reading, and relevant quotes where appropriate. With web search on, Claude can also fetch a specific page when a user pastes its URL. How search gets switched on depends on the account: - In the standard chat, a user turns on "Web search" from the "+" menu, and can turn it off again. - On Team and Enterprise plans, an Owner or Primary Owner must first enable web search for the whole workspace before members can use it. - Anthropic says that in "the new Claude experience" there is no toggle, and Claude searches the web when it helps. Developers building on Anthropic's API get a separate web search tool. Its documentation says Claude decides when to search, the response includes cited sources, and developers can restrict results with `allowed_domains` or exclude sites with `blocked_domains`. It says Claude tends to search for things that change, including "information about specific organizations, people, or products that might have changed." So an app built on Claude may never be able to see your site, whatever you do. That limit is set by whoever built the app. The practical point: the training route is slow and outside your control, while the search route reads your pages as they are now. That makes search the route your site changes can affect. ## What Anthropic documents about its crawlers Anthropic's help page on web crawling lists three bots. This table states only what that page says. | Bot | What Anthropic says it does | What blocking it does, per Anthropic | | --- | --- | --- | | ClaudeBot | Collects web content that could potentially contribute to training its models | Signals that your site's future material should be excluded from training datasets | | Claude-User | Accesses websites when individual users ask Claude questions | Stops Claude retrieving your content in response to a user query, which "may reduce your site's visibility for user-directed web search" | | Claude-SearchBot | Navigates the web to improve the quality of search results | Stops Anthropic indexing your content for search, which "may reduce your site's visibility and accuracy in user search results" | The same page says Anthropic's bots honour robots.txt directives, support the non-standard `Crawl-delay` extension, and will not try to bypass CAPTCHAs. You have to add rules for every subdomain you want to cover. Anthropic also warns that blocking its IP addresses "may not work correctly" as an opt-out, because that stops it reading your robots.txt. It publishes a list of its crawler IP addresses so you can check whether a visit really came from Anthropic. The split means you can choose separately. A site can block ClaudeBot (keeping future content out of training) and still allow Claude-SearchBot and Claude-User (staying reachable for search). Whether that trade suits you is your decision. Anthropic does not say that blocking ClaudeBot affects search. On robots.txt itself, RFC 9309 is the standard. It says crawlers "are requested to honor" the rules, and that the rules "are not a form of access authorization." Rules sit in groups, one per user agent, and a crawler follows the group that matches its name. If no group matches, it follows the `*` group, if there is one. So a rule for ClaudeBot says nothing about Claude-SearchBot, but a blanket `*` rule can catch all three. For writing the rules, see our guide to [AI crawlers and robots.txt](/ai-crawlers-robots-txt/). ## What you can control on your own site **Crawler access (documented).** This is the one lever Anthropic describes. Check your robots.txt for a `User-agent: *` group with `Disallow: /`, and for old rules that block bots by name. Either can shut out Claude-SearchBot and Claude-User without anyone meaning to. Check every subdomain too, as Anthropic's instructions apply per subdomain. Our [crawler access](/guide/#crawler-access) section covers the basics for all the main AI crawlers. **Page content (recommended).** Anthropic publishes nothing about what makes a page more likely to be cited, so the points below are our advice. 1. Describe your business in one plain sentence on your home and about pages: what you do, for whom, and where. 2. Use the same business name and service names everywhere you appear, on your own site and elsewhere online. 3. Put the direct answer to a buyer's question near the top of the page that covers it, then add the detail. 4. Make sure the important text is in the page's HTML and does not depend on scripts or images to appear. None of these guarantees a mention. They make your pages easier to find, read and quote. ## Which parts are research and which are practice The main research here is GEO: Generative Engine Optimization by Aggarwal and colleagues, accepted to KDD 2024. The authors built GEO-bench, a benchmark of 10,000 queries. They tested content changes on a generative engine they set up themselves, which answered from the top five Google results for each query, and repeated the tests on Perplexity.ai. Adding citations, quotations from relevant sources and statistics improved visibility in their tests. They also found the effect varied by domain. Its limits matter. Claude was never tested, the engine was the researchers' own, and visibility was measured with their own metrics. Treat it as evidence that content changes can affect citations in that setting. It does not show that the same edits will work in Claude. Everything else on this page, including the content advice above and the checking method below, is practice. It is sensible practice, and it is labelled as advice. ## Why "ranking in Claude" is the wrong mental model A search engine returns an ordered list, and you can track your position in it. Claude writes an answer, and nothing in Anthropic's documentation describes a ranked list you could climb. The same question can produce different wording and different names from one run to the next, and the API documentation says developers can steer, through their system prompt, how readily Claude searches. "Claude SEO" is a fair label for this work, and we use it that way. It is the same job as for other assistants: be reachable, be clear, and be worth quoting. If you have read our guides on [how to rank in ChatGPT](/how-to-rank-in-chatgpt/) or [Perplexity SEO](/perplexity-seo/), the approach will look familiar. What changes is the crawler names and how search is switched on. ## How to check whether Claude names you The most direct check is to ask Claude yourself, in a fixed way. This method is our recommendation. 1. Write a short list of questions a real buyer would ask when choosing a business like yours (we suggest eight to twelve). Fix the wording and do not change it. 2. Ask each question in a fresh chat with web search on. Where your account allows it, also ask with web search off, so you can see what Claude says from training alone. 3. Log the date, the model, whether search ran, whether you were named, whether your site was cited, which competitors appeared, and how Claude described you. 4. Run the full set several times, as one answer proves little when answers vary. 5. Repeat on a regular schedule and after any change to your site or robots.txt, then compare the logs. For a fuller method across assistants, see [how to measure AI visibility](/how-to-measure-ai-visibility/) and our notes on [measuring progress](/guide/#measurement). ## What to do in what order 1. **Access first.** Read Anthropic's crawler page, then check robots.txt on every subdomain so Claude-SearchBot and Claude-User are not blocked by accident. Decide on ClaudeBot separately. 2. **Then the page basics.** Write a plain description of what you do, use consistent names, and give direct answers to buyers' questions. 3. **Then a baseline.** Run your fixed question set and log the results before you change anything else, so you have something to compare against. 4. **Then repeat.** Rerun the set after changes and over time. A wider [AI visibility audit](/ai-visibility-audit/) looks at the same questions across several assistants, and [how to improve AI visibility](/how-to-improve-ai-visibility/) picks up from there. Doing it in this order means you never spend time polishing a page that Claude's search bots cannot reach. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you have run a few buyer questions by hand, it tracks over time whether Claude and other assistants name and cite your company for those questions. ## Questions people ask ### Does Claude search the web? Yes. Anthropic's help centre says Claude can search the live web and that every search response includes citations. In the standard chat, users turn it on from the "+" menu, and Team and Enterprise owners must first enable it for the workspace. Anthropic says the new Claude experience has no toggle and searches when it helps. ### How do I get my website cited by Claude? Nobody can guarantee it. The documented step is to let Claude-SearchBot and Claude-User reach your pages through robots.txt. Clear, consistent pages that answer buyers' questions directly are our recommendation. ### Which Anthropic crawlers should I allow in robots.txt? Anthropic lists three: ClaudeBot (training), Claude-User (fetching pages when users ask questions) and Claude-SearchBot (improving search results). Anthropic says blocking either of the last two may reduce your visibility in Claude's search answers. Blocking ClaudeBot signals that your future content should be kept out of training. ### Is there such a thing as Claude SEO? As a label for this work, yes. It is not a separate discipline: Anthropic documents no ranking factors, so the job is the same as for other assistants. Be reachable by the right crawlers, be clear and be worth quoting. ### How can I check whether Claude mentions my business? Ask Claude a fixed set of buyer questions in fresh chats, several times, with web search on (and off where you can). Log whether you are named or cited and who else appears. Repeat over time and after any site change. ## Sources - [Claude Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Claude Help Center: Enable and use web search](https://support.claude.com/en/articles/10684626-enabling-and-using-web-search) - [Claude Help Center: How up-to-date is Claude's training data?](https://support.claude.com/en/articles/8114494-how-up-to-date-is-claude-s-training-data) - [Claude Platform Docs: Web search tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: How do you show up in Gemini? What is documented, what is research, and what is guesswork Canonical: https://whatisgeo.app/how-to-show-up-in-gemini/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Nobody can guarantee a mention in Gemini. Google documents that AI Overviews and AI Mode draw on Search and that the Gemini app can ground answers in the Search index, so being indexable is the main lever. Start by working out which Gemini your customers actually use, because Google documents each one in a different place and the controls differ. In every case Google describes, the web content Gemini uses comes through Google Search, so the work starts with ordinary crawlability and indexing. This page labels each claim. **Documented** means Google publishes it. **Researched** means a paper tested it, under conditions we state. **Recommended** is our own advice. ## Which Gemini do you mean? The app, AI Overviews, AI Mode and the API "Gemini" covers several things, and the most useful first step is deciding which one matters to you. | Surface | Where people meet it | Where Google documents it | |---|---|---| | Gemini app | gemini.google.com and the mobile apps | Gemini Apps Help Centre | | AI Overviews | A summary on a Google results page | Google Search Central | | AI Mode | A conversational mode inside Google Search | Google Search Central | | Gemini API with Search grounding | Apps that developers build on Gemini | Google AI for Developers | The most common mistake is treating all four as one channel. Google's own Search documentation says AI Mode and AI Overviews "may use different models and techniques, so the set of responses and links they show will vary", and those two are the closest pair. If your buyers mostly meet you on a results page, read our guides to [ranking in Google AI Overviews](/how-to-rank-in-google-ai-overviews/) and [Google AI Mode SEO](/google-ai-mode-seo/). This page covers what applies across all four. ## What Google documents about how Gemini finds web content **AI Overviews and AI Mode (documented).** Google Search Central's page on AI features says that to be eligible as a supporting link, "a page must be indexed and eligible to be shown in Google Search with a snippet", and that "there are no additional technical requirements." It also says specific optimisation "isn't required" and that the usual SEO fundamentals still apply. Google adds that both features may use "query fan-out", running several related searches across subtopics to build one answer. **The Gemini app (documented).** Google's crawler documentation describes grounding in the Gemini apps as "providing content from the Google Search index to the model at prompt time." The Gemini Apps Help Centre says the apps "may provide links to sources and related content", which can include public websites. It also says that not every response includes sources. If there is no Sources button, the app gave no links for that answer. **The Gemini API (documented).** Grounding with Google Search is a tool a developer switches on. Google's developer docs say the model decides whether a search would improve the answer, runs one or more queries if so, and returns a response with inline citations to its sources. This explains how Gemini can use Search. It says nothing about how individual pages get chosen. Across all three, the web content Google describes comes from its Search index. That much is documented. Exactly how pages get picked from that index for any given answer is not published. ## What you control: crawling, indexing and Google-Extended You control two separate settings, and they are easy to mix up. Our home guide's section on [crawler access](/guide/#crawler-access) explains the general idea. **Googlebot controls Search, including its AI features (documented).** Google says AI is "integral to how Search functions", so robots.txt rules for Googlebot are how you control crawling for Search. To limit what Search shows from a page, it points to `nosnippet`, `data-nosnippet`, `max-snippet` and `noindex`. Because snippet eligibility is a stated condition, a page with `nosnippet` or `noindex` cannot be a supporting link in AI Overviews or AI Mode. **Google-Extended controls Gemini training and app grounding (documented).** Google describes Google-Extended as "a standalone product token" for robots.txt. It does not crawl on its own and has no separate user agent. It decides whether content Google crawls from your site may be used to train future Gemini models (for the Gemini apps and the Vertex AI API for Gemini) and for grounding in the Gemini apps and in Grounding with Google Search on Vertex AI. Google states that it "does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search." In practice: - Blocking Google-Extended leaves AI Overviews and AI Mode alone, because Google ties those to Googlebot and Search. - Blocking Google-Extended can keep your content out of grounded answers in the Gemini app. If you want to appear there, check you have not blocked it by accident, for example through a blanket rule copied from an "AI bots" list. Google updates these pages, so check the current wording before you change robots.txt. Our guide to [AI crawlers and robots.txt](/ai-crawlers-robots-txt/) covers the other assistants' crawlers. ## What research says about getting cited by generative engines The most cited study is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024 (researched). The authors built a benchmark of 10,000 queries. For each query they gave a language model the text of the top five Google results and asked it to write an answer. Then they rewrote individual sources in different ways and measured how visible each source became in the answer. They report gains of up to 40% on their own visibility measures. Adding citations, quotations from relevant sources and statistics did especially well. They also found that what worked varied by domain. They repeated the methods on Perplexity.ai as a check on a live product. Bear its limits in mind: - **It did not test Gemini,** the Gemini app, AI Overviews or AI Mode. - **Most results came from a setup the authors built** and scored on their own metrics. A live product that changes without notice may behave differently. - **It measured visibility inside answers,** not traffic, leads or sales. The fair conclusion is that how a page is written can change how much of it a generative engine uses, and that the effect depends on the subject. It does not tell you that any given edit will get you named in Gemini. ## What is sensible practice (and labelled as such) These are our recommendations. They follow from the documentation above, and none of them comes with a guarantee. 1. **Make sure key pages are indexable and snippet-eligible.** This is the one step with documentation behind it. Use the URL Inspection tool in Search Console to check. 2. **Check robots.txt for Googlebot and Google-Extended separately.** Decide on purpose what each one should be allowed to do. 3. **Answer the question early and in plain words.** Put the fact, price or condition near the top of the page, where a model summarising it has less to misread. 4. **Keep business details consistent.** Use the same name, description, location and offer on your site, your Google Business Profile and other listings. Google's AI features page itself advises keeping Business Profile and Merchant Center information up to date. 5. **Test with real buyer questions,** the ones customers actually ask before they buy. This is mostly good Search practice. That is to be expected, because the documented route into Gemini runs through Search. ## What nobody can promise Nobody can promise you a mention in Gemini, and that includes Google, agencies, tool vendors and us. Google does not document any inclusion guarantee for any Gemini surface. The research does not offer one either. Nobody can honestly promise a timeframe or a percentage uplift. Answers also vary. Google says AI Overviews often do not appear at all, and that AI Mode and AI Overviews can show different links for the same question. One check therefore tells you very little. If someone sells you numbered steps, ask which Google page or paper each step comes from. ## How to check whether Gemini names you, step by step You can run this yourself. It shows where you stand today and predicts nothing. 1. **List 10 to 20 real buyer questions** in your customers' own words. 2. **Pick one surface per test.** Use the Gemini app, a Google search (for AI Overviews), or AI Mode. Do not mix results from different surfaces. 3. **Ask each question and record the result.** Note whether you are named, whether you are linked (in the app, open the Sources panel) and which sites appear instead. 4. **Repeat on other days.** Answers shift between runs, so look for patterns. 5. **Check the basics on any page you expected to see.** Is it indexed? Is it eligible for a snippet? Is Google-Extended set the way you intended? 6. **Use Search Console for AI Overviews and AI Mode.** Google says clicks and impressions from these features count in the Performance report under the "Web" search type. They are combined with ordinary results there, not broken out separately. For a fuller method that covers other assistants too, read [how to measure AI visibility](/how-to-measure-ai-visibility/). ## Gemini compared with ChatGPT and Perplexity for visibility The documented difference is where the content comes from. For Gemini, Google's documentation points back to its own Search index every time. ChatGPT and Perplexity use their own crawlers and their own documentation, so a robots.txt rule or a fix that matters for one may do nothing for the others. We cover them separately in [how to rank in ChatGPT](/how-to-rank-in-chatgpt/) and our Perplexity guide. We have no evidence for ranking one above another as a channel. The useful comparison is your own: run the same buyer questions in each and see where you are named. ## Questions The common questions about Gemini visibility are answered below. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you have checked a few questions by hand, it tracks whether Gemini and other assistants name and cite your company for your buyers' questions over time. ## Questions people ask ### How do I get my website to show up in Gemini? Nobody can guarantee it. Google's stated requirement for AI Overviews and AI Mode is that a page is indexed and eligible to show with a snippet, with no additional technical requirements. Beyond that, clear pages, consistent business details and testing with real buyer questions are our recommendations. ### Does blocking Google-Extended stop me appearing in Gemini or AI Overviews? Google says Google-Extended does not affect inclusion or ranking in Google Search, and points to Googlebot as the robots.txt control for AI features in Search. It does govern whether Google may use your content to train future Gemini models and to ground answers in the Gemini apps, so blocking it can affect the Gemini app. ### Is Gemini SEO different from Google SEO? For AI Overviews and AI Mode, Google says the same foundational SEO practices apply and no specific optimisation is required. Anything beyond that is someone's opinion, including ours. ### Do I need to submit my site to Gemini? None of the Google documentation we checked describes a separate submission step. For AI features in Search, the stated condition is being indexed and eligible for a snippet. ### Is Gemini better than ChatGPT for SEO? We have no evidence to rank them. They reach web content in different ways, so test each one with your own buyer questions and compare the answers you actually get. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google: List of Google's common crawlers (Google-Extended)](https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers) - [Google AI for Developers: Grounding with Google Search](https://ai.google.dev/gemini-api/docs/google-search) - [Gemini Apps Help: View related sources from Gemini Apps](https://support.google.com/gemini/answer/14143489) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: How do you track AI Overviews? What Google reports, what tools estimate, and what you can check yourself Canonical: https://whatisgeo.app/how-to-track-ai-overviews/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Google counts AI Overview clicks inside your normal Search Console web data, with no separate report. Tracking means combining that, your analytics and sampled checks, and none of them gives a complete count. Search Console already counts AI Overview clicks and impressions, and GA4 already files those visits under Organic Search. Neither lets you pull them out on their own, so the missing piece is a record of when an AI Overview appears and whether it links to you, from your own checks or from a tool's sample. ## What you can and cannot track There are three separate questions, and each method answers a different one. | Question | Best view | Main limit | | --- | --- | --- | | Does an AI Overview appear for my searches, and am I cited? | Manual checks or a third-party tracker | Only covers the queries, places and moments sampled | | Are people clicking through to my site? | Search Console and GA4 | AI Overview clicks are mixed in with other Google organic clicks | | How often does this happen across my market? | A third-party tracker | Its keyword list and settings decide the answer | None of these gives you a complete count of every AI Overview a real person saw. For how this fits with measuring AI answers more widely, see [measuring progress](/guide/#measurement). ## What Google itself reports about AI features Google's Search Central page "AI features and your website" says that sites appearing in AI features such as AI Overviews and AI Mode are included in the overall search traffic in Search Console. They are reported in the Performance report, within the "Web" search type. Search Console Help ("What are impressions, position, and clicks?") sets out how they are counted: - **Clicks.** Clicking a link to an external page in an AI Overview counts as a click. - **Impressions.** For AI Overviews, Google says "Standard impression rules apply." - **Position.** "An AI Overview occupies a single position in search results," and every link inside it gets that same position. - **AI Mode.** A click on an external link in AI Mode counts as a click, and a follow-up question is treated as a new query. - **Search Labs.** Search Console does not include data from experiments in Search Labs. Neither page describes a filter, search type or search appearance option that isolates AI Overviews, and the Performance report help lists none. Our reading of that documentation: you can see AI Overview traffic in Search Console, but you cannot separate it from the rest of your web search data. Google also says AI Mode and AI Overviews may use different models and techniques, so the responses and links they show will vary. Both may use "query fan-out", running related searches across subtopics to build a response. That is one documented reason a page can be cited for a search where it does not rank in the classic results. ## Tracking visibility: is your site cited in AI Overviews? Google provides no feed that tells you a page was cited. You find out by looking, either yourself or through a tool, and both are sampling. On variation, Google's documentation says responses differ between AI Mode and AI Overviews. It does not list every other cause. In practice (our observation, not a Google statement) the same search can show a different AI Overview, or none, depending on the wording, location, device and the time you check. A single screenshot is one data point, so log every check with its date, device and location. ## Tracking traffic: what Search Console and GA4 can and cannot show **Search Console.** Use the Performance report with the "Web" search type and filter to the queries and pages you care about. Clicks and impressions from AI Overviews are already in those numbers. Because all links in an AI Overview share one position, a change in average position on a query can mean an AI Overview has appeared or gone, as well as ordinary ranking movement. Every figure you read there is a combined one. **GA4.** Google Analytics Help defines the Organic Search channel as visits from non-ad links in organic search results, "including Google's AI Overviews and AI Mode". The AI Assistant channel, which covers referrals from tools such as ChatGPT, Gemini and Copilot, explicitly excludes AI Overviews and AI Mode. Google's guide to using Search Console and Analytics together identifies Google search visits by session source "google" and session medium "organic". So in GA4, an AI Overview click arrives as ordinary Google organic traffic, and the documentation describes no way to split it out. Any "AI Overviews traffic" figure from GA4 alone is an estimate. Do not expect Search Console and GA4 to agree exactly. Google says clicks and sessions are calculated differently, and that what matters is that the general trends follow the same pattern. **Text fragment links.** Some practitioners point out that AI Overview links can carry a text fragment (the `#:~:text=` part of a URL) and suggest using it to tag that traffic. Google does not document this as a tracking method. MDN explains that browsers strip the fragment directive from the URL during loading so that page scripts cannot read it directly, which means a standard analytics tag will not see it. Treat any setup built on this as an unconfirmed technique and test it on your own site before relying on it. ## Manual checks done properly Manual checks are slow, but they show exactly what a searcher saw. This routine is our recommendation: 1. Pick a fixed list of queries that matter to your business, and keep it stable so results stay comparable. 2. Search in a clean, signed-out browser window to reduce personalisation. 3. Record the date, device and location for each check. 4. Note whether an AI Overview appeared, whether your site was linked, and which other sites were. 5. Repeat on a schedule rather than whenever you feel curious. A spreadsheet is enough. The limit is coverage: you only learn about the queries you thought to type, and a competitor may be cited on many searches you never tried. ## What third-party trackers do, and how to judge one Several rank trackers now report on AI Overviews. On their own feature pages: - **Semrush** says its rank tracker "tracks AI Overviews" as well as AI Mode and other AI tools, and that it tracks visibility across devices and locations. - **Ahrefs** says Rank Tracker monitors 19 SERP features, "including AI Overviews", on mobile and desktop, with locations from country down to ZIP code. - **SE Ranking** says its tools check visibility across AI Overviews, AI Mode, ChatGPT and Gemini, with an AI Results Tracker showing daily positions. Those are the vendors' own descriptions. Prices change and differ by plan, so check each vendor's pricing page on the day you look. Whatever the tool, it runs the queries you (or it) chose, from the locations and devices it simulates, at the times it checks. Its output is a sample of results and never a census of what searchers saw. That is how we recommend reading any of them, whichever vendor it is. When comparing, ask: 1. Can you choose the queries, and how many are included? 2. Which locations and devices does it simulate? 3. How often does it re-check? 4. Does it show the captured result so you can verify it? 5. Does it tell a linked citation apart from a brand mention in the text? A tool that shows its working is easier to trust than one that hands you a single score. Many small sites will not need one at all. ## A simple weekly routine This is a suggested routine, small enough to keep doing: 1. In Search Console, note movement for your key pages and queries in the "Web" search type. 2. In GA4, check Organic Search landing pages for the same pages. Look at direction, not decimals. 3. Run your manual query list and log the results. 4. If you use a tracker, compare its sample with your manual log, and look closer where they disagree. 5. Write one line on what changed. Weekly is usually enough. Day-to-day movement in AI results is easy to over-read. ## What not to read into the numbers - **A tracker percentage describes its sample.** "Cited on a share of tracked keywords" says nothing about searches it did not run. - **A traffic dip does not prove an AI Overview caused it.** Seasonality, ranking changes and changes to your own site all move organic visits. - **A missing citation is a single observation.** You may not have checked the right search, place or moment. - **Click-loss statistics need scrutiny.** Many published figures come from vendors. Ask who ran the study, on which queries, in which country and over what period, and treat a vendor's number as that vendor's claim. For how AI answers are put together in the first place, see [how AI answers happen](/guide/#how-it-works). Common questions about GEO are answered in [questions](/guide/#questions), and the rest of our guides are in [all guides](/guides/). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Search Console only covers Google, so if you also want to know whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions, it tracks that. ## Questions people ask ### Can you see AI Overviews in Google Search Console? Yes, but not on their own. Google says traffic from AI features such as AI Overviews and AI Mode is included in the Performance report under the "Web" search type, and its documentation describes no filter that separates it out. ### How do you track traffic from AI Overviews in GA4? You cannot isolate it in GA4 alone. Google Analytics Help says the Organic Search channel includes Google's AI Overviews and AI Mode, so those visits arrive as ordinary Google organic traffic. ### Do AI Overview clicks show up as organic search traffic? Yes. Search Console Help says clicking an external link in an AI Overview counts as a click, and GA4 puts AI Overviews and AI Mode in the Organic Search channel. They are excluded from GA4's AI Assistant channel. ### Do I need a paid tool to track AI Overviews? No. Search Console, GA4 and a regular manual log cover a lot. A paid tracker adds scale across more queries, but what it gives you is still a sample of results, so judge it on how openly it shows its working. ### Why do I see different AI Overviews from other people for the same search? Google says AI Overviews and AI Mode may use different models and techniques, so the responses and links vary. In our experience wording, location, device and timing also make a difference, which is why a dated log tells you more than one screenshot. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Search Console Help: What are impressions, position, and clicks?](https://support.google.com/webmasters/answer/7042828) - [Search Console Help: Performance report (Search results)](https://support.google.com/webmasters/answer/7576553) - [Google Analytics Help: Default channel group](https://support.google.com/analytics/answer/9756891) - [Google Search Central: Using Search Console and Google Analytics data for SEO](https://developers.google.com/search/docs/monitor-debug/google-analytics-search-console) - [MDN: Text fragments](https://developer.mozilla.org/en-US/docs/Web/URI/Reference/Fragment/Text_fragments) - [Semrush: Keyword rank tracker (vendor feature page)](https://www.semrush.com/features/keyword-rank-tracker/) - [Ahrefs: Rank Tracker (vendor feature page)](https://ahrefs.com/rank-tracker) - [SE Ranking: Rank Tracker (vendor feature page)](https://seranking.com/position-tracking.html) --- # Guide: How do you turn off Google AI Overviews? What Google allows, and what only hides them Canonical: https://whatisgeo.app/how-to-turn-off-google-ai-overviews/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Google says AI Overviews cannot be turned off. Its Web filter shows results without them, a browser shortcut can make that view your default, and extensions only hide them on your own screen. Google's own help page is plain about it: AI Overviews are "a core Google Search feature" and "Features cannot be turned off." You can still get results without them, one search at a time with the Web filter or by default with a browser shortcut, and each method below is labelled by how far it depends on Google leaving things as they are. ## Can you turn off AI Overviews at all? Not with a setting. Google's Search Help article on AI Overviews compares them to knowledge panels and says features cannot be turned off. The same paragraph gives the alternative: select the Web filter after you search, which "displays only text-based links without features like AI Overviews." Every method on this page carries one of three labels: - **Google documented:** Google describes it in its own help pages. - **Unofficial workaround:** it works because of how Google's pages behave today, and Google can change that without notice. - **Our reading:** our own reasoning or recommendation, with no source behind it. One caveat. The AI Overviews help article has a notice saying Google is updating its Search services settings and the article may not match what you see. Google's page on those new settings covers search history and personalised recommendations. It does not describe a switch for AI Overviews. If Google ever adds one, the AI Overviews help article is where to look. ## Why the old Search Labs and SGE opt-out advice is out of date Many older guides tell you to open Search Labs and turn the experiment off. That advice dates from the Search Generative Experience (SGE), which Google introduced in 2023 as "an experiment in Search Labs". In May 2024 Google said people had used AI Overviews "billions of times through our experiment in Search Labs" and started rolling them out to everyone in the US. From then on, AI Overviews were part of normal Search and no longer an opt-in test. Labs still exists. Google's help page calls "AI in Search" in Labs a "testbed" for new AI features and experiments, and gives the steps to opt out: sign in (not in Incognito), open a new tab, select Labs, then Manage, and toggle off "AI in Search". That opts you out of the experiments. Google does not say it removes AI Overviews, and the same page says Search features cannot be turned off. If you have turned Labs off and still see AI Overviews, nothing is wrong. That setting just controls something else. Label: Google documented, for Labs experiments only. ## Methods Google documents: the Web filter Google's "Refine Google searches" help page lists the filters that appear below the search bar after you search, Web among them, and says the Web filter "contains text-based links to websites." The AI Overviews help page confirms that this view leaves AI Overviews out. 1. Search as normal. 2. Choose **Web** from the filters below the search bar. 3. If Web is not showing, select "All" to see the other filters. Google says which filters appear, and in what order, depends on the query. There are two limits. The filter only applies to the search you are on, so you have to pick it again next time unless you set up a shortcut (see the next section). It also drops other features, such as knowledge panels, so some searches will look sparse. Label: Google documented. ## Browser-level workarounds: making the Web view your default When you choose the Web filter, the address of the results page changes. If you save that address in your browser as a custom search engine, every search from the address bar opens in the Web view. The browser makers document how to add a custom search engine. Google does not document the part of the address that selects the Web view. That part is the parameter `udm=14`. Tedium, an independent newsletter, wrote it up in May 2024 and suggested `https://www.google.com/search?q=%s&udm=14` as the search address. Tedium's udm14.com warns that if it stops working, "Google sometimes introduces random breaking changes." You can also find the current address yourself: search for an easy word, choose Web, copy the address bar, and replace your word with `%s`. Chrome's and Mozilla's help pages both describe this copy-and-replace method. ### Chrome on a computer Google Chrome Help gives these steps: More, Settings, Search engine, then Manage search engines and site search. Next to "Site search", select Add and fill in a name, a shortcut and the address with `%s`. Then, next to the new entry, select More and Make default. Chrome notes that at work or school your administrator may control the search engine. ### Firefox on a computer Mozilla Support says you can add a custom search engine in Settings from Firefox version 140. Earlier versions need an add-on. In Firefox's search settings, use Add under Search Shortcuts, or Add search engine under Additional search engines (which one you see depends on your version). Paste in the address with `%s`, add a keyword if you want one, and select Add Engine. You can then choose it as your default. ### Safari on Mac and iPhone Apple's help page for Safari on Mac covers choosing a search engine under Safari, Settings, Search. It describes no way to add a custom search address, so Apple documents no way to make the Web view your default. Tedium mentions third-party Safari extensions. We have not checked them and do not list them here. ### Chrome on Android and iPhone Google Chrome Help for Android and for iPhone and iPad describes picking a default from a list, where "recently visited search engines are added as options." Neither page describes adding a custom address. On a phone, the documented route is the Web filter, chosen on each search. Label: unofficial workaround. The browser steps are documented by the browser makers, but the Google parameter they depend on is not documented by Google. ## Extensions and filter lists: what they do and their risks There are two kinds. - **Search engine add-ons** set up the shortcut for you. One example is the "udm14" add-on on Mozilla's Firefox add-ons site. Its page says it adds `udm=14` to Google searches so you can make it your default, and that add-ons of this type do not work in Firefox on Android. It is third-party, and we name it only as an example of the type, not as a recommendation. - **Hiding extensions and content-blocker rules** take the AI Overview off the page after it loads. They depend on how Google builds its results page, so when Google changes that, a rule can stop working and you get no warning. We do not name one. Our reading: either kind can read the pages you visit once you grant it permission, so check what an extension asks for before you install it. Label: unofficial workaround. ## Methods compared | Method | What it does | Documented by Google? | How long it lasts | Works on mobile? | |---|---|---|---|---| | Web filter | Shows text links only, no AI Overview | Yes | One search at a time | Yes, where the filter appears | | Custom search engine with `udm=14` | Opens the Web view by default | Browser steps yes, the parameter no | Until Google changes how its addresses work | Not on the Chrome or Safari mobile pages we checked | | Search engine add-on | Sets up the same shortcut for you | No | Same as above, plus the add-on's upkeep | The udm14 add-on says not on Firefox for Android | | Hiding extension or filter rule | Removes the block from your screen | No | Until Google changes its page | Depends on the browser | | Labs "AI in Search" opt-out | Leaves Google's Labs experiments | Yes | Permanent | Yes | Read the Labs row closely: it is documented and it lasts, but it does not remove AI Overviews. ## What hiding AI Overviews does not change This section is our reading. Every method above changes what you see and nothing else. The Web filter applies to your own search, a shortcut lives in your browser, and an extension edits the page on your device. People searching the same thing still get whatever Google shows them. If what you want is for AI Overviews to go away for everyone, none of these will do it. For background on the feature itself, see [what Google AI Overviews are](/what-are-google-ai-overviews/) and our overview of [how AI answers are put together](/guide/#how-it-works). ## If you run a website: what this means for you If you hide AI Overviews in your own browser, your view of where your pages appear will be skewed (our reading). Other searchers may still see AI Overviews above your results. Check from a normal, unfiltered search, and measure over time from outside your own setup. [How to track AI Overviews](/how-to-track-ai-overviews/) covers what Google reports and what you can check yourself. As a site owner, the documented controls you have work on your pages, not on anyone's screen. Google Search Central says that to limit the information shown from your pages in Search you can use `nosnippet`, `data-nosnippet` or `max-snippet`. To work on the other side, appearing in AI Overviews, see [how to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/) and, for the conversational version of Search, [Google AI Mode and SEO](/google-ai-mode-seo/). ## Questions people ask ### Can you turn off AI Overviews in Google Search? No. Google's Search Help page calls AI Overviews a core Search feature and says features cannot be turned off. It points to the Web filter as the way to see results without them. ### Does the Web filter remove AI Overviews? Yes, for the search you apply it to. Google says the Web filter displays only text-based links without features like AI Overviews. You have to choose it again on each search unless you set up a browser shortcut. ### Is the Search Labs opt-out still a way to turn off AI Overviews? No. Turning off "AI in Search" in Labs opts you out of Google's experiments. AI Overviews left the Labs experiment in May 2024 and became part of normal Search, and Google says Search features cannot be turned off. ### How do I hide AI Overviews on iPhone, Android or Safari? Use the Web filter after each search. The Chrome and Safari help pages we read for phones, and Apple's Safari page for Mac, only describe picking a search engine from a list, with no option to add a custom search address. ### Do AI Overview extensions or blocking rules stop working? They can. They are third-party tools that depend on how Google builds its results page or its URLs, and Google can change either without notice. ## Sources - [Google Search Help: AI Overviews in Google Search](https://support.google.com/websearch/answer/14901683?hl=en) - [Google Search Help: Refine Google searches (filters, including Web)](https://support.google.com/websearch/answer/2466433?hl=en) - [Google Search Help: Search Services History and Personalized Recommendations](https://support.google.com/websearch/answer/17025248?hl=en) - [Google: Supercharging Search with generative AI (SGE in Search Labs, 2023)](https://blog.google/products/search/generative-ai-search/) - [Google: Generative AI in Search, Google I/O 2024 (AI Overviews rollout)](https://blog.google/products/search/generative-ai-google-search-may-2024/) - [Google Chrome Help: Set default search engine and site search shortcuts (computer)](https://support.google.com/chrome/answer/95426?hl=en) - [Google Chrome Help: Set default search engine (Android)](https://support.google.com/chrome/answer/95426?hl=en&co=GENIE.Platform%3DAndroid) - [Google Chrome Help: Set default search engine (iPhone and iPad)](https://support.google.com/chrome/answer/95426?hl=en&co=GENIE.Platform%3DiOS) - [Mozilla Support: Add or remove a search engine in Firefox](https://support.mozilla.org/en-US/kb/add-or-remove-search-engine-firefox) - [Apple Support: Change Search settings in Safari on Mac](https://support.apple.com/guide/safari/search-sfria1042d31/mac) - [Tedium: How I made Google's Web view my default search (third party, May 2024)](https://tedium.co/2024/05/17/google-web-search-make-default/) - [udm14.com (third-party shortcut site by Tedium)](https://udm14.com/) - [Firefox Add-ons: udm14 (third-party add-on)](https://addons.mozilla.org/en-US/firefox/addon/udm14/) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) --- # Guide: How do you write content for AI search? What is documented, what is research, and what is folklore Canonical: https://whatisgeo.app/how-to-write-content-for-ai-search/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Answer the question plainly near the top, make each section stand alone and source your facts. Google says its AI features need no special optimisation, and nobody can guarantee a citation. No platform documents a special writing style for AI search. What Google and OpenAI publish is short and familiar, the main study tested rewrites under lab conditions, and most of the confident rules online have nothing public behind them. This page sorts the advice into three groups. **Documented** means a platform says it. **Research** means a named study found it, under conditions we state. **Recommended** is our own advice, labelled as ours. ## What AI search systems do with your writing Google documents that AI Overviews and AI Mode may use a "query fan-out" technique: they run several related searches across subtopics and data sources, then build one response that can link to a wider set of pages than a classic results page. OpenAI says its OAI-SearchBot crawler is used to surface websites in ChatGPT's search features. The documentation does not say how passages are picked. Our working model, which is a recommendation and not a documented fact, is that one part of your page may be used next to parts of other pages. So each part should make sense on its own, say something specific and be easy to check. The home guide's [principles](/guide/#principles) set out the same idea for the whole site. ## What the platforms document about content Google's Search Central post of 21 May 2025, by John Mueller, lists what it advises for AI experiences: - Focus on "unique, non-commodity content" that is useful to visitors. - Provide a good page experience. - Make sure Googlebot is not blocked, that pages return a successful HTTP status and that the content is indexable. - Use preview controls (nosnippet, data-nosnippet, max-snippet, noindex) to decide what appears, "including in our AI formats". - Make sure "all the content in your markup is also visible on your web page". - Support the text with high-quality images and videos. Google's AI features documentation says a page must be indexed and eligible to show in Search with a snippet, that "there are no additional technical requirements", and that no special optimisations are needed for AI Overviews or AI Mode. It adds that there is no special schema.org markup you need to add, and that indexing and serving are not guaranteed. More on markup is in [schema markup for AI](/schema-markup-for-ai/). Two documented points matter to writers because they can stop good sentences from ever being used: - **Snippet controls.** Google's robots meta tag documentation says nosnippet also stops the content being used as a direct input for AI Overviews and AI Mode, and max-snippet limits how much of it may be used there. data-nosnippet keeps marked parts of a page out of snippets. - **Crawler access.** OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. The setup is covered in [AI crawlers and robots.txt](/ai-crawlers-robots-txt/). Google's guidance on helpful, people-first content suggests judging a page by "Who, How, and Why". It says E-E-A-T is not a specific ranking factor and that trust is the most important of its four parts. Two of its self-assessment questions apply at sentence level: does the main heading give "a descriptive, helpful summary of the content", and does the content provide "original information, reporting, research, or analysis"? ## What the research found, and its limits The main study is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024. The authors tested nine ways of rewriting a source: making it more authoritative, adding statistics, keyword stuffing, citing sources, adding quotations, making it easier to understand, improving fluency, using unique words and adding technical terms. They built a benchmark called GEO-bench of about 10,000 queries drawn from nine sources. They ran most tests on a simulated engine built on gpt-3.5-turbo, using the top five Google results for each query, and checked a 200-query subset on Perplexity.ai. The abstract reports that GEO "can boost visibility by up to 40%" in generative engine responses. Adding citations, quotations and statistics did best. Keyword stuffing gave little or no improvement. Read those results with their conditions attached: - Visibility was measured with the authors' own metrics: a word count weighted by citation position, and impression scores rated by GPT-3.5. - The sources were supplied to the engine, so the study tests what happens once a page is already among the sources. It does not test whether a page gets found. - Results varied by domain, and the authors say methods "may need to adapt over time" as engines change. - The engines tested in 2023 and 2024 are not the products running today. The fair reading: adding sourced facts and quotations is supported by one study under lab conditions. That supports the sourcing advice below, and it is not a guarantee. ## Write the answer first, then the detail **Recommended.** Put the plain answer in the first sentence or two of the page, and again at the top of each section. Then give the reasoning, the conditions and the exceptions. If the honest answer is "it depends", name what it depends on straight away. This serves a reader who skims, and it gives any system that reads one passage a complete answer. Nobody has published evidence that it raises citation rates, so treat it as good writing. ## Make each section stand on its own **Recommended.** Write every section as if it might be the only part someone reads. 1. Name things in full the first time they appear in a section ("Acme Heating's annual boiler service" rather than "it" or "this option"). 2. Make headings say what the section says. "Pricing" tells a reader less than "What a boiler service costs and what it includes". Google's own self-assessment questions ask for descriptive headings. 3. Do not make a section depend on a definition three screens up. Repeat the short version. If a passage only makes sense after reading everything above it, it is fragile, for a person arriving from a search result and for a system pulling one passage. ## Be specific, and source your claims **Recommended**, and loosely backed by the research above. Replace vague claims with specific ones a reader can check. Say where a figure came from, date facts that change, and update them when they go stale. If you cannot source a claim, soften it or cut it. This fits Google's people-first framing, where trust matters most. It also guards against the worst outcome: a confident, out-of-date sentence that gets repeated. ## Match the question people actually ask **Recommended.** Write headings and opening sentences in your buyers' words, not your internal ones. If customers ask "how much is a boiler service for a flat", a heading called "Service packages" misses them. Look at the questions in your enquiries, reviews and sales calls. To see which of those questions AI assistants already name you for, see [how to measure AI visibility](/how-to-measure-ai-visibility/). ## Tips you will see that have no public evidence These turn up on vendor and agency pages near the top of search results, written as if they were rules. We found no platform documentation or published study behind any of them: | Claim | What we found | |---|---| | Answers must be 40 to 60 words | No documented or researched length. A rule of thumb. | | Add an FAQ block to every page | Not a documented requirement. Not tested in the GEO study. | | Write in "chunks" of a set size | No published source for any ideal length. | | You need special schema for AI | Google says no special schema.org markup is needed. | A short answer near the top is good writing. The exact number is what has no evidence behind it. Be wary of anyone who sells these as certainties. ## A paragraph before and after This is an invented example for a fictional heating firm. Before: > Our team has been designed with a wide range of customers in mind, and we believe that keeping your home warm doesn't have to be complicated. Many customers find the experience smooth, and there are various options available depending on what you need. Nothing there can be lifted or checked. It has no answer, no name and no fact. After: > Acme Heating's annual boiler service covers a safety check, a clean of the main parts and a written report. A Gas Safe registered engineer does the work. Prices for gas, oil and LPG boilers are on our prices page, and we update them each April. What changed: the business and service are named, the answer comes first, each claim can be checked, and the page says when the prices change. We added no keywords, no target length and no markup trick. ## What writing cannot fix Good prose will not rescue a page that is blocked, not indexed or set to nosnippet. It will not make a commodity page unique, and it cannot guarantee a citation, because nobody can. Check access first, then the writing. The platform detail is in [the Google AI Overviews guide](/how-to-rank-in-google-ai-overviews/). If you are drafting with AI tools, read [whether AI content is good for SEO](/is-ai-content-good-for-seo/). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you have rewritten a page, you can check whether ChatGPT, Claude, Gemini and Perplexity name and cite you for your buyers' questions. ## Questions people ask ### How do you structure content for AI search? We recommend opening with the direct answer, using headings that describe what each section says, and writing each section so it makes sense on its own. None of this is a documented requirement. Google says its AI features need no special optimisation beyond being indexed and eligible for a snippet. ### Should answers be 40 to 60 words? We found no platform documentation or published study that sets an answer length. The 40 to 60 word figure is a rule of thumb repeated on vendor and agency pages. A short, plain answer near the top is good writing, but no exact word count is required. ### Does writing in a question-and-answer format help you get cited? No platform documentation we read says it does, and the GEO study did not test it. Use a question-and-answer layout where readers really do arrive with separate questions, and keep any FAQ markup matched to text that is visible on the page. ### Do I need to write differently for AI Overviews, ChatGPT and Perplexity? Nothing we read says so. Google states there are no special optimisations needed for AI Overviews or AI Mode, and OpenAI's documented requirement is crawler access for OAI-SearchBot. One clear page written for the reader is the sensible default. ### Can I use AI to write content for AI search? You can, as long as the result is accurate, specific and useful to the reader. Google's people-first guidance asks you to consider who made the content, how and why, so check every fact and add what only you know before you publish. ## Sources - [Google Search Central Blog: Top ways to ensure your content performs well in Google's AI experiences on Search (John Mueller, 21 May 2025)](https://developers.google.com/search/blog/2025/05/succeeding-in-ai-search) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central: Robots meta tag, data-nosnippet and X-Robots-Tag specifications](https://developers.google.com/search/docs/crawling-indexing/robots-meta-tag) - [Google Search Central: Creating helpful, reliable, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [OpenAI: Overview of OpenAI crawlers](https://developers.openai.com/api/docs/bots) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: How to claim your business in ChatGPT, and what Business Hub's insights mean for GEO Canonical: https://whatisgeo.app/claim-business-in-chatgpt/ Publisher: Adapt Progress Evolve Limited Updated: 2026-11-10 > Business Hub lets eligible businesses claim their ChatGPT profile, edit its details and see insights. UK online businesses with a website may be able to claim now; UK local shops and Business Agents have to wait. Whether you can claim today depends on what kind of business you run, and Business Agents are not open to any UK business yet. Below: the claim steps OpenAI documents, what you can change, how to read the new numbers for generative engine optimisation (GEO), and what is still unknown. ## What OpenAI launched On 10 November 2026 OpenAI announced "Businesses in ChatGPT". It is in beta. OpenAI puts ChatGPT at 1.2 billion weekly users. That is OpenAI's own figure, and it is why the launch matters: many people ask questions there, and some of the answers name businesses. The launch has three parts: 1. **Claiming.** You claim your business in Business Hub, verify it and keep its details current. 2. **Insights.** Aggregate numbers on how people engage with your business in ChatGPT. 3. **Business Agents.** Eligible businesses can build an agent that answers people's questions about their offerings in a separate, clearly labelled conversation, distinct from ChatGPT's own answers. For most businesses the profile is the part that matters. When a business appears in an answer, people can click through to its ChatGPT profile, which shows business details and a link to its website. The profile sits between the answer and your site, so it is worth knowing what it says. ## Can a UK business claim? OpenAI says any eligible business located where ChatGPT is available can claim, and that eligibility depends on factors including industry, location and its verification and integrity checks. Its help article adds two limits that decide things for UK businesses today: | Your business | What OpenAI's help article says | |---|---| | Local store or restaurant | "Local stores and restaurants can currently claim only an existing U.S. listing." A UK high street shop or restaurant cannot claim yet. | | Online business with a website | "If you run an eligible online business, you may see an option to add it." The product page says claiming "is initially available to eligible businesses with a website". | | Shopify merchant | You can connect your store from Business Hub or the Shopify admin to verify. Eligible Shopify products are already in ChatGPT through Shopify Catalog. | You need to own the business or be authorised to represent it. You do not need an ads account or a ChatGPT Enterprise agreement. ## How to claim your business These are the steps in OpenAI's help article: 1. Sign in to ChatGPT and open **Business Hub** from the sidebar (or go to chatgpt.com/business-hub). Search for your business by name. 2. Select the listing that matches your business and confirm which ChatGPT account will manage it. 3. Choose one of the verification methods offered and follow the prompts. The methods offered vary by business. OpenAI lists four: - **Business phone:** a code by text message or automated call. - **Business email:** a code sent to the email address associated with the business. - **Website domain:** add the TXT record ChatGPT shows you to your domain's DNS, then select **Verify domain**. If someone else runs your domain, ask them to add it. - **Shopify store:** connect and log in to the store. OpenAI then reviews the claim and emails you its decision or a request for more information. It says review time varies. **The second check: Persona.** An approved claim is not the end. To publish changes to your details, connect a product feed or view insights, you also need identity verification through Persona, which may ask for government-issued ID, business registration information and proof you are authorised. OpenAI says the business name, address and any registration or tax number must match your official records, so use your Companies House details exactly. The blue check on your profile appears only after Persona approval. One caution: OpenAI says a profile created through Persona is published automatically when approved, and only unpublished draft profiles can be deleted. Check the profile before you submit. ## What a verified profile lets you change Once both checks are approved, OpenAI says you can update your website, business description, product or service category, logo, delivery policy and return policy, and add a cover photo and other optional details. OpenAI reviews and publishes changes, and they can take time to appear in a listing or an answer. It also says ChatGPT may use other sources when answering questions about your business, so the profile is one input among several. If you sell products, you can connect a product feed with prices and availability, either a CSV upload or a hosted file that follows OpenAI's product feed specification, or a Shopify feed. OpenAI says connecting a feed does not guarantee ChatGPT recommends your products. ## What Business Hub insights show Insights cover aggregate activity over the past 28 days. You do not see individual conversations; OpenAI says it does not share users' conversations or Business Agent chats with businesses. - **Mentions.** For an online business, a mention counts when ChatGPT references the business's website. The brand name alone does not count. - **Website clicks.** Clicks through to your site. - **Discovery topics.** Common themes across conversations where your website is cited. - **Profile views.** Shown on the dashboard in OpenAI's screenshot. Insights appear only when there are enough conversations or interactions to meet OpenAI's thresholds, and it has not published those thresholds. A small business may find very little at first. That tells you about volume, not quality. ## Using the insights for GEO GEO is the work of making a business easy for AI assistants to find, understand and cite (the guide's [principles](/guide/#principles) cover the approach). For the wider method, see [how to rank in ChatGPT](/how-to-rank-in-chatgpt/); for how ChatGPT finds and cites pages, see [ChatGPT search](/chatgpt-search/). The suggestions below are our recommendations for reading these numbers, not OpenAI guidance. **Start with discovery topics.** They are the closest thing to a view of why your site gets cited. Ask two questions of each: is this what you want to be known for, and do your pages say enough about it? A topic you care about that your site covers thinly is the clearest thing to fix, because your own pages are what you control. **Remember what counts as a mention.** Because an online business is only counted when ChatGPT references its website, a low mention count can mean your site is rarely cited even if your name comes up. That points back at your pages, not your profile. **Read mentions and clicks together, and over time.** Each view is a rolling 28 days, so compare one period with the next. If mentions rise and clicks do not, people see your site referenced but do not follow the link; look at the page it points to. If clicks rise, check what that page does with the visit. **Keep your website as the source of truth.** Clear pages that answer buyers' questions, with facts that agree everywhere they appear, are what the profile and the citations rest on. To track this with a method rather than by feel, [how to measure AI visibility](/how-to-measure-ai-visibility/) sets out one you can run yourself, and the guide's [measurement section](/guide/#measurement) has the short version. OpenAI says plainly that claiming does not guarantee higher placement or sales. Treat the insights as a record of what happened, not a forecast. ## What UK businesses cannot do yet - **Claim a local listing.** Local stores and restaurants can currently claim only an existing US listing. - **Business Agents.** These are for eligible ecommerce and online service businesses in the US with a website. - **Sponsored Business Agents.** Businesses can sponsor their agent and use it as the destination for ChatGPT Ads, with sponsored placements labelled. That needs an agent, so it is US-only too. OpenAI says it plans to expand Businesses in ChatGPT and Business Agents to more markets and categories over time. It has given no dates. ## What we do not know yet OpenAI's announcement, product page and help article leave these open: - whether a particular UK online business is eligible, until you try (eligibility depends on industry, location and OpenAI's checks); - whether claiming will ever cost anything (none of the three pages mentions a price); - how long claim and Persona reviews take; - the thresholds insights need before they appear; - when local listings and Business Agents open outside the US. Where a UK small business stands today: if you run an online business with a website, look for it in Business Hub and claim it, using your official company details for Persona. If you run a local shop or restaurant, you have to wait. Either way, your website remains the thing ChatGPT cites, and we will update this guide as OpenAI publishes more. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Business Hub only shows activity once it passes OpenAI's thresholds; this tracks whether ChatGPT and other assistants name and cite your company for your buyers' questions either way. ## Questions people ask ### Can a UK business claim its profile in ChatGPT? Possibly, depending on the kind of business. OpenAI says local stores and restaurants can currently claim only an existing US listing, while an eligible online business may see an option to add itself. Eligibility also depends on industry, location and OpenAI's checks. ### Do I need an ads account or ChatGPT Enterprise to claim? No. OpenAI's help article says you can claim a business you own or are authorised to represent without an Ads account or an Enterprise agreement. Claiming, publishing an agent and advertising are separate. ### Will claiming my business make ChatGPT recommend it more? No promise of that. OpenAI says claiming does not guarantee higher placement or sales, and that ChatGPT keeps making independent recommendations. ### Can I see what people asked ChatGPT about my business? No. OpenAI does not share users' conversations or Business Agent chats with businesses. You see aggregate activity over the past 28 days, and only once there is enough activity to pass its thresholds. ### When will Business Agents be available in the UK? OpenAI has not given a date. For now they are for eligible ecommerce and online service businesses in the US with a website. ## Sources - [OpenAI: Introducing Businesses in ChatGPT (10 November 2026)](https://openai.com/index/introducing-businesses-chatgpt/) - [OpenAI Help Center: Setting up your business on ChatGPT (checked 10 November 2026)](https://help.openai.com/en/articles/20001635) - [OpenAI: Businesses in ChatGPT, product page and FAQ (checked 10 November 2026)](https://openai.com/business/businesses-in-chatgpt) --- # Guide: How to improve your AI visibility: what you can influence and what you cannot Canonical: https://whatisgeo.app/how-to-improve-ai-visibility/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > AI visibility is how often AI answers name, cite and correctly describe your business. You can improve your odds with crawler access, clear pages and consistent details. Nobody can guarantee inclusion. Start by measuring, because AI answers vary from one run to the next and you need a baseline to compare against. Then check crawler access, make your pages easy to quote, remove contradictions about your business across the web, and re-check the same way. Each step below says what kind of claim it rests on: a platform's own documentation, published research, or sensible practice. | Step | What you do | What it rests on | |---|---|---| | 1. Baseline | Ask assistants buyer questions repeatedly and log the results | Sensible practice | | 2. Access | Allow the search crawlers you want in robots.txt and at your CDN or firewall | Platform documentation (Google, OpenAI, Perplexity, Anthropic) | | 3. Quotable pages | Direct answer first, specific sentences, one question per page | Sensible practice, partly backed by Google's guidance and one research paper | | 4. Consistency | Same name, location, services and facts everywhere | Sensible practice, partly backed by Google's guidance | | 5. Re-check | Repeat the baseline and compare patterns | Sensible practice | ## What AI visibility means (and what it does not guarantee) Three things have to happen. The assistant names you. It cites you, with a link or source attribution a reader can follow. And it describes you correctly: right services, right location, right facts. Miss any one and you have a problem. Being named but described wrongly is arguably worse than not being named at all. Visibility is not a ranking. There is no fixed results page to sit at the top of, and no guaranteed slot for a given question. Google's own documentation says AI Overviews and AI Mode may use different models and techniques, so the responses and links they show will vary, and that AI Overviews only appear when Google's systems judge them useful, so they often don't trigger at all. What you can realistically aim for is better odds, measured properly. That is why the order of work matters: measure first, change things second, check again third. For the wider picture of how answers are assembled, see [how AI answers happen](/guide/#how-it-works). ## Step 1: Find out where you stand today This is a recommendation from practice, not a platform rule, and it is the step most people skip. Write down the questions a buyer would really ask. Leave out your brand name and describe the problem you solve: "Who does X near me?" or "What should I look for in a Y supplier?" Put those questions to the assistants your customers are likely to use and record what comes back. Do it more than once. Answers vary between runs, so a single query tells you very little. Ask each question several times, ideally on different days, and log three things each time: were you named, were you cited, and was the description correct? A plain spreadsheet does the job, with one row per question per run. You are looking for patterns, not a single verdict. Keep the log, because without it every later claim of improvement is guesswork. If you are setting this up for the first time, [the 30-day plan](/guide/#getting-started) covers the groundwork. ## Step 2: Make sure the right crawlers can reach you If a crawler cannot read your pages, it cannot use them. The major providers document their own crawlers, and they are not interchangeable: | Provider | Crawler | What the provider says it does | |---|---|---| | OpenAI | OAI-SearchBot | Surfaces sites in ChatGPT's search features. Sites that block it will not be shown in ChatGPT search answers, though they can still appear as navigational links. | | OpenAI | GPTBot | Crawls content that may be used to train OpenAI's foundation models. | | OpenAI | ChatGPT-User | Visits pages when a user asks something. OpenAI says robots.txt rules may not apply, and it is not used to decide search inclusion. | | Perplexity | PerplexityBot | Surfaces and links sites in Perplexity search results. Perplexity says it is not used to crawl content for AI foundation models. | | Perplexity | Perplexity-User | Fetches pages for user questions. Perplexity says it generally ignores robots.txt, because a user requested the fetch. | | Anthropic | ClaudeBot | Collects content that may contribute to model training. | | Anthropic | Claude-User | Fetches pages when people ask Claude questions. Anthropic says blocking it may reduce your visibility for user-directed web search. | | Anthropic | Claude-SearchBot | Indexes content to improve Claude's search results. Anthropic says blocking it may reduce your visibility in search results. | Because search and training crawlers are separate, opting out is not all-or-nothing. OpenAI states plainly that you can allow OAI-SearchBot to appear in search results while disallowing GPTBot. Read each provider's page before you edit anything, since the names and behaviour are theirs to define and the pages get updated. Robots.txt is only part of it. Google lists making sure crawling is allowed "by any CDN or hosting infrastructure" as a best practice, and Perplexity's documentation explains how to allow its bots through a web application firewall. A firewall rule set up years ago can block crawlers that your robots.txt welcomes. Two cautions. First, robots.txt is a request, not a lock. RFC 9309, the standard that defines it, says its rules "are not a form of access authorization". If something needs to be private, put it behind a login. Second, Google works differently. For AI Overviews and AI Mode, Google says the same foundational SEO best practices apply. To be shown as a supporting link, a page must be indexed and eligible to appear in Search with a snippet, and "there are no additional technical requirements". Google also says robots.txt rules for Googlebot are the control for Search, including its AI features. Google-Extended is a separate control for training and grounding in some of Google's other systems, so it is not the switch for AI Overviews. More detail is in [crawler access](/guide/#crawler-access). ## Step 3: Make pages easy to quote Most of this step is our recommendation. Google's guidance supports parts of it: it asks for helpful, reliable, people-first content and for important content to be available as text. Put the direct answer first. If a page exists to answer a question, answer it in the opening lines in plain words, before the context and caveats. A system looking for a passage to use will find one, and a person skimming will too. Write sentences that still make sense when quoted on their own. "We do it differently" means nothing by itself. A sentence that names the service, the place and the fact can be quoted. Specific beats clever. Keep pages focused: one page, one question, one clear answer. A page that sprawls across six topics is harder to quote accurately. None of this is exotic. It is good writing for readers, and it happens to suit systems that summarise and cite. The thinking behind it is set out in [the principles](/guide/#principles). ## Step 4: Be described consistently across the web Assistants do not only read your site. They also draw on what others say about you, so your business can be described in several places at once, and those descriptions can disagree. Keep your business name, location, services and key facts consistent across your own site and your profiles. Spell the name the same way, use the same address format and list the same services. If you move or drop a service, update it everywhere, not just on your home page. This is mostly a recommendation from practice, though Google's AI features page does list two related checks: keep your Business Profile and Merchant Center information up to date, and make sure your structured data matches the visible text on the page. It costs little and reduces the chance of an assistant stitching a muddled description together from conflicting sources. You cannot make a third party say anything, but you can make sure everything you control says the same thing. ## What the research says, and its limits The most cited study is Aggarwal et al., *GEO: Generative Engine Optimization*, accepted to KDD 2024. The authors built GEO-bench, a benchmark of 10,000 queries, and used a generative engine of their own that answered each query from the top five Google results, roughly mimicking engines like Perplexity. They then rewrote source pages in different ways and measured visibility with their own metrics. The rewrites that added citations, quotations from relevant sources and statistics did best. They report improvements of up to 40% on their own visibility measure, and they also ran the same methods on Perplexity.ai. Keyword stuffing, a classic SEO tactic, gave little to no improvement. They also found that what works varies by domain. Those results come from a specific benchmark, a specific engine design and metrics the authors defined, with some scoring done by an earlier language model (GPT-3.5). Today's assistants work differently and keep changing. The paper is a reasonable argument for clear, well-supported writing and for measuring what happens. It does not predict what any one change will do for your site. You will also see vendors quoting big percentage lifts, sometimes borrowed from this paper without its conditions. If a figure comes from a vendor, it is the vendor's claim, and you should weigh it that way. ## What you cannot control You cannot control whether an assistant includes you in a given answer, how it words its description, or which sources it prefers on a given day. You cannot make its output repeatable: the same question can return different answers on different runs. None of the provider documentation listed on this page describes a way to pay or submit your way into the answer itself. OpenAI's crawler page does mention ads on ChatGPT, but that is advertising, which is a separate thing from the answer. You also cannot control what third parties say about you, or when providers change their crawlers and policies. Re-read the provider pages from time to time rather than relying on an old screenshot. What you can control is in Steps 2 to 4: access, clarity and consistency. That is less than the hype suggests, but it is real. ## How to tell whether it is working Go back to your baseline log. Run the same questions again, the same number of times, and record the same three things: named, cited, described correctly. Compare patterns, not single answers. One run where you are named proves little, and so does one run where you are missing. If, across many runs, you are named more often, cited more reliably or described more accurately than before, that is worth taking seriously. If nothing changes, that tells you something too. For Google specifically, traffic from AI Overviews and AI Mode is included in Search Console's Performance report, counted within the "Web" search type. Remember that changes take time to be picked up: Google says crawling can take anywhere from several days to several months, depending on how often its systems decide a page needs refreshing. Be careful about giving your changes the credit. Assistants change on their own, so a better result might come from your work or from an update you never saw. Note what you changed and when, and don't declare success early. There is more on this in [measuring progress](/guide/#measurement). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you have a baseline routine, this tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your business for buyers' questions over time. - [Hilyt](https://hilyt.it/): If assistants describe your business inconsistently, an open AI-readable profile gives them one clear statement of who you are and what you do. ## Questions people ask ### How do I improve my business's visibility in ChatGPT and AI Overviews? Record a baseline, check the right crawlers can reach you, make pages clear and quotable, keep your business details consistent, then re-check the same way. For ChatGPT search, OpenAI recommends allowing OAI-SearchBot in robots.txt. For AI Overviews and AI Mode, Google says a page must be indexed and eligible for a snippet, with no additional technical requirements. ### Is AI visibility different from SEO rankings? They overlap but they are not the same thing. Google says the same foundational SEO best practices apply to its AI features, so good SEO groundwork still counts there. AI visibility asks whether generated answers name, cite and correctly describe you, and those answers vary rather than sitting in a fixed position. ### Can I pay or submit my site to get into AI answers? None of the crawler and AI features documentation we read from Google, OpenAI, Perplexity or Anthropic describes a way to pay or submit your way into the answer itself. OpenAI's crawler page does mention ads on ChatGPT, which are a separate paid placement. Google's documentation says no special optimisation is required for AI Overviews and AI Mode. ### How do I check whether AI mentions my business? Ask the assistants your customers use the questions a buyer would ask, several times each on different days, and record whether you are named, cited and described correctly. Answers vary between runs, so compare patterns across many runs. For Google, AI Overviews and AI Mode traffic is counted within Search Console's Performance report under the Web search type. ### How long does it take to improve AI visibility? Nobody can honestly promise a timeframe. Google says crawling can take anywhere from several days to several months, depending on how often its systems decide a page needs refreshing, and assistants change independently of anything you do. Judge progress against your own baseline. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO paper, full text (HTML v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: Is AI content good for SEO? What Google says, and what actually matters Canonical: https://whatisgeo.app/is-ai-content-good-for-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Google judges content on quality, whatever tool made it, and treats mass-produced pages that add no value as spam. AI content can work for SEO when it is fact-checked, original and useful. The deciding question is why a page exists and what a reader gets from it. Google documents this clearly, and it also tells you to fact-check AI output before publishing, a step most flat "yes" or "no" answers leave out. ## Three kinds of claim, kept apart Most answers to this question are opinions. A vendor says yes because it sells the tools, and a forum thread says no because someone's traffic fell. Neither is evidence. This page sticks to three kinds of claim and labels each one: 1. **What Google documents.** The firmest ground, because Google sets the rules for its own Search. 2. **Research.** This page cites no study and uses no traffic or ranking statistics, because we only use research we have read and can put in context. 3. **Sensible practice.** Where something is our recommendation, we say so. Keeping these apart clears up most of the confusion. "Google penalises AI" mixes a real spam policy with rumour. "AI is fine, publish at will" ignores that same policy. Both make the same mistake: they treat the production method as the whole story. ## What Google documents about AI-generated content Google sets out its position in a Search Central blog post from February 2023, "Google Search's guidance about AI-generated content", and in a documentation page on using generative AI content. The key lines: - **Quality is the focus, however content is made.** The blog post says Google focuses on "the quality of content, rather than how content is produced". - **Appropriate use is allowed.** From its FAQ: "Appropriate use of AI or automation is not against our guidelines." - **Manipulation is spam.** Using automation, including AI, to generate content with the primary purpose of manipulating search rankings is, in Google's words, "a violation of our spam policies". - **AI earns nothing special.** "Using AI doesn't give content any special gains. It's just content." If it is useful, helpful, original and shows aspects of E-E-A-T, "it might do well in Search. If it doesn't, it might not." - **Review is expected.** The documentation page says generative models "don't retrieve facts, but predict a likely sequence of words", so their outputs may contain inaccuracies, and that "it is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing". That review covers metadata too: title elements, meta descriptions, structured data and image alt text. That last point matters. Fact-checking AI output is something Google itself tells you to do, in writing. ## Where AI content gets into trouble: scaled content abuse Google's spam policies define scaled content abuse as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users", typically "large amounts of unoriginal content that provides little to no value to users, no matter how it's created". The listed examples include: - using generative AI tools or similar tools to generate many pages without adding value for users - scraping feeds, search results or other content to generate many pages, including through automated synonymising, translating or other obfuscation - stitching together content from different pages without adding value - creating multiple sites to hide the scaled nature of the content - creating many pages that make little or no sense to a reader but contain search keywords Notice the phrase "no matter how it's created". The policy also covers human-written pages made in bulk for the same purpose. AI comes up so often because it makes producing many pages cheap and fast, which is the pattern the policy describes. The same spam policies page also counts attempts "to manipulate generative AI responses in Google Search" as spam. There is a quieter risk below the spam line. Google's people-first guidance lists warning signs of "search engine-first content", including "Are you using extensive automation to produce content on many topics?" and "Are you mainly summarizing what others have to say without adding much value?" A page can stay clear of every spam policy and still be too generic to earn anything. ## What helpful content looks like, and where AI drafts fall short Google's guidance on helpful, reliable, people-first content gives self-assessment questions that read like an editor's checklist. Among them: does the content provide original information, reporting, research or analysis? Does it have any easily verified factual errors? Is it written or reviewed by an expert or enthusiast who demonstrably knows the topic? Does it show first-hand expertise, "for example, expertise that comes from having actually used a product or service, or visiting a place"? The same page explains E-E-A-T (experience, expertise, authoritativeness and trustworthiness). It says trust is the most important of the four, and that E-E-A-T "itself isn't a specific ranking factor". It also suggests asking "Who, How, and Why" about your content, and describes the "Why" as "perhaps the most important question". Content made primarily to help people aligns with what Google's systems seek to reward; content made primarily to attract search visits does not. Our view, offered as practice and not as Google's rule: those questions are exactly where unedited AI drafts struggle. - A model has not used the product, visited the place or run the test. It can describe those things fluently, and fluent is not the same as accurate. - Drafts tend to summarise what already exists, so they rarely add anything new. - Nobody is accountable for a draft until a named person puts their name to it. None of that makes AI unusable. It means the human review stage is essential. ## Do you need to disclose AI use? The Search Central pages we read do not require a disclosure on ordinary web pages. They recommend one where it makes sense. Google's FAQ says AI or automation disclosures "are useful for content where someone might think 'How was this created?'" and suggests adding them "when it would be reasonably expected". For content substantially generated by automation, the people-first page adds questions such as whether that use is "self-evident to visitors through disclosures or in other ways". The FAQ also says giving AI an author byline "is probably not the best way" to make AI's role clear to readers. Shopping has a firmer rule, and it is written down. Google's generative AI page says that for ecommerce sites, Merchant Center requires AI-generated images to carry IPTC `TrainedAlgorithmicMedia` metadata, and requires AI-generated product titles and descriptions to be labelled as AI-generated. If you sell through Merchant Center, read its own policy. Our recommendation: where readers would reasonably wonder how a page was made, a short note on how AI was used and who reviewed it costs little. ## A review checklist before you publish The first item is Google's documented advice. The rest are our recommendation; treat them as the minimum. 1. **Fact-check every claim**, plus the title, meta description, structured data and alt text. If you cannot source a claim, cut it. 2. **Add first-hand detail.** What did you or your team actually see, run, measure or get wrong? 3. **Name the expertise.** Say who wrote or reviewed the page and why they know the topic. 4. **Edit for originality.** Ask what this page says that the pages already ranking do not. If the answer is nothing, rethink it. 5. **Check the "why".** Would you publish this page for your existing readers if search engines did not exist? 6. **Do not bulk-publish without review.** No batch of pages should go live until a person has read each one. The [GEO principles](/guide/#principles) in our main guide apply the same thinking to search and AI answers together. ## Does this change for AI Overviews, ChatGPT and Perplexity? For Google, no. Google's page on AI features says the same foundational SEO best practices apply to AI Overviews and AI Mode as to Search overall, including creating helpful, reliable, people-first content. To be shown as a supporting link, a page must be indexed and eligible to appear in Search with a snippet, and "there are no additional technical requirements". It also says "specific optimization isn't required". Our guide to [Google AI Overviews](/how-to-rank-in-google-ai-overviews/) goes into more detail. For ChatGPT, Perplexity and other assistants, this page quotes no documentation on AI-written content, and we will not invent any. As practice and no more: accurate, clearly sourced pages with a named author seem sensible to aim for if you want to be cited, and nothing guarantees a mention. For how the two disciplines overlap, see [GEO vs SEO](/guide/#geo-vs-seo) and our guide to [LLM SEO](/llm-seo/). ## What nobody can tell you Nobody can promise that an AI-assisted page will rank, or that it will not. Google publishes its policies and its intent, never a formula, so treat any claim of certainty as a sales line, whoever makes it. What you can do is publish carefully, then watch your own results and adjust. The [measurement section](/guide/#measurement) of our main guide explains how to track progress in search and in AI answers. ## From the same studio These are made by the studio that publishes this guide. - [ScriptGrain](https://scriptgrain.com/): If you draft with AI, ScriptGrain produces first drafts in your own measured voice for you to fact-check and edit as this page recommends. ## Questions people ask ### Does Google penalise AI-generated content? Not for being AI-generated. Google says appropriate use of AI or automation is not against its guidelines, and that using automation to generate content primarily to manipulate search rankings breaks its spam policies. ### Is ChatGPT content good for SEO? It can be, if the published page is helpful, accurate and original. Google does not single out any tool, and says using AI gives content no special gains either way. Our recommendation is to treat the output as a first draft, then fact-check and edit it. ### Can you publish AI content without editing it? Google's guidance says it is critical to manually fact-check and review all AI-generated content for accuracy before publishing. Publishing many unreviewed pages that add no value matches Google's own example of scaled content abuse. ### What is scaled content abuse? It is a Google spam policy covering many pages generated for the primary purpose of manipulating search rankings and not helping users. It applies no matter how the content is created, so bulk human-written pages count too. ### Do you need to disclose that content is AI-generated? The Search Central pages do not require it for ordinary web pages; Google suggests adding a disclosure where readers would reasonably expect one. Merchant Center is stricter: Google says AI-generated images need IPTC metadata and AI-generated product titles and descriptions must be labelled. ## Sources - [Google Search Central: Google Search's guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central: Spam policies for Google web search (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: Creating helpful, reliable, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) --- # Guide: Is there a Chrome extension to hide Google AI Overviews? What they do and what to check first Canonical: https://whatisgeo.app/hide-google-ai-overviews-extension-chrome/ Publisher: Adapt Progress Evolve Limited Updated: 2026-11-10 > Yes. Several Chrome extensions hide the AI Overview on your own screen, but none can switch it off, because Google says it cannot be turned off. Check site access, privacy declarations and update dates first. Every one of these extensions works the same way. It edits the Google results page inside your browser after the page arrives, so the AI Overview is still produced and simply not shown to you. Before you install one, ask three questions: what can it access, what does its developer declare about your data, and is anyone keeping it up to date? ## What a hiding extension actually does A hiding extension runs inside Chrome on Google's search pages. When a results page loads, it finds the part of the page that holds the AI Overview and hides it. Nothing changes on Google's side. The listings say as much if you read past the name. The "Google AI Overview Hider" listing uses "Disable AI Overview" and "Turn Off AI Overview" in its marketing, then explains in its own "How it works" section that it works "by hiding them from the Search Page". It also says that switching the extension off reloads the page "to show the original content". Whatever a store listing calls it, hiding is the only thing an extension can do here. ## What Google says about turning AI Overviews off Google's Search Help page calls AI Overviews "a core Google Search feature, like knowledge panels" and says "Features cannot be turned off." The same paragraph gives Google's alternative: select the Web filter after you search, which "displays only text-based links without features like AI Overviews." Google offers no extension or setting that removes AI Overviews. Our guide on [how to turn off Google AI Overviews](/how-to-turn-off-google-ai-overviews/) covers every route, including why the old Search Labs opt-out does not do it. ## Extensions listed in the Chrome Web Store Searching the Chrome Web Store for "Google AI Overview Hider" on 19 October 2026 returned more than ten extensions offering to hide or remove AI Overviews. The three below are examples, in alphabetical order, with what each listing said that day. We have not installed or tested any of them, and this is not a ranking or a recommendation. | Listing | What the listing says it does | Updated (per listing) | Privacy section (per listing) | |---|---|---|---| | AI Overview Hider for Google (soitis.dev) | "Hides the AI Overview section in Google search results" | 5 April 2026 | Developer says it will not collect or use your data | | Google AI Overview Hider | Hides AI Overviews and "AI Mode" labels with an on/off toggle, "bringing the top organic search results back to the top of the page" | 23 April 2026 | Developer says it will not collect or use your data, and links a privacy policy | | Hide Google AI Overview (ternera) | Hides the AI Overview from Google results; says it is open source and links its code | 30 December 2025 | Developer says it will not collect or use your data | Firefox has its own equivalents. The "Hide Google AI Overviews" add-on on Mozilla's add-ons site describes itself as "Hide annoying Google AI Overviews", and says it hides the AI Overviews at the top of search results, the "People also ask" box, the AI-generated sections of "Things to know" and the AI Mode tab. Its listing is also the clearest example of what can go wrong, covered below. We leave out user counts and star ratings. They change daily, several of these listings have only a handful of ratings, and neither tells you whether an extension is safe. Check the listing yourself on the day you install. ## How to check an extension before you install it Chrome and the Web Store give you three things to check, and each takes a minute. 1. **Who publishes it and when it was last updated.** Every Chrome Web Store listing has a Details section showing the version, the "Updated" date and the developer's contact details. Some also carry a badge: the soitis.dev listing says it was "Created by the owner of the listed website" and that the publisher "has a good record with no history of violations." Our reading: for a page Google keeps changing, a recent update date matters more than a high rating. 2. **The Privacy section.** Chrome's developer documentation says developers "must disclose how your extension collects and uses user data" in the Web Store's privacy fields, and certify that they comply with its limited use policy. That disclosure is what appears under "Privacy" on the listing. It is the developer's own declaration, so read it as their statement of what the extension does with your data. 3. **What it asks for, and what you allow afterwards.** Chrome Web Store Help says some extensions tell you at install time if they "need certain permissions or data", and advises: "Make sure you only approve extensions that you trust." After installing, open More, then Extensions, then Manage extensions, and select Details on the extension. Next to "Allow this extension to read and change all your data on websites you visit", you can set its site access to On select, On specific sites or On all sites. Our reading: a hider only needs Google's search pages, so limiting it to the Google site you use is a sensible default. Firefox shows required permissions on the listing page itself. The Hide Google AI Overviews listing, for example, asks for access to Google's search pages across Google's country domains, and its description says it "runs only on Google search pages, collects no data, and makes no network requests". ## What can go wrong Hiding depends on how Google builds its results page. When Google changes the page, the extension can stop finding the AI Overview until its developer ships a fix, and nothing warns you. You just see the AI Overview again. The Firefox Hide Google AI Overviews listing shows this happening. When we checked it on 19 October 2026, it carried an "April Update" saying the developers were "aware the extension no longer functions as of late March", had updated the add-on and were awaiting review from Mozilla. When we checked again on 10 November 2026, that notice was no longer in the description. The listing showed version 1.7.2, whose release note says it "Fixes search results appearing squeezed into a narrow column on Google's new layout (follow-up to 1.7.1)". Two more risks, both our reading: - **Abandoned extensions.** If nobody updates an extension, it will break the next time Google changes its page, and it can stay listed in the store after it stops working. - **Access to your searches.** An extension that can read and change Google's search pages can see what you search for. The Privacy section is the developer's statement about what happens to that data, which is why it is worth reading. Chrome Web Store Help also says Chrome may disable extensions that do not keep up with new requirements. ## Options that need no extension **The Web filter.** Search as normal, then choose Web from the filters under the search bar. Google documents this option. The catch is that you have to choose it again on each search. **A custom search engine in Chrome.** This makes the Web view your default. Chrome Help gives these steps: 1. Open More, then Settings, then Search engine, then Manage search engines and site search. 2. Next to "Site search", select Add. 3. Give it a name and a shortcut. For the address, run any search, choose Web, copy the address from the address bar and replace your search word with `%s`. 4. Select Add, then select More next to the new entry and choose Make default. The part of that address that selects the Web view is `udm=14`, which Tedium's udm14.com offers as a shortcut to an "AI-Free" Google search. Google does not document that parameter, and the FAQ on udm14.com warns that "Google sometimes introduces random breaking changes." Neither route needs permissions or updates from a third party. The [off-switch guide](/how-to-turn-off-google-ai-overviews/) has the Firefox and phone versions. ## Does hiding AI Overviews change what searchers see or what you should do as a site owner This section is our reading. A hiding extension changes your screen and nothing else. Other people searching the same thing still see whatever Google shows them, and Google still generates the AI Overview. If you run a website, hiding the block on your own screen changes nothing about whether your site is cited in AI Overviews. It can, however, skew your view of your own search results, so check from a browser without the extension. [How to track AI Overviews](/how-to-track-ai-overviews/) covers what you can measure. If your concern is traffic, see [do Google AI Overviews reduce clicks](/do-google-ai-overviews-reduce-clicks/). If you want to appear in them, see [how to rank in Google AI Overviews](/how-to-rank-in-google-ai-overviews/). For background, see [what Google AI Overviews are](/what-are-google-ai-overviews/) and [how AI answers happen](/guide/#how-it-works). ## Questions people ask ### Is there a Chrome extension that hides Google AI Overviews? Yes. When we searched the Chrome Web Store on 19 October 2026, more than ten extensions offered to hide AI Overviews, including listings named "Google AI Overview Hider" and "AI Overview Hider for Google". None of them is made by Google. ### Does an extension turn AI Overviews off, or just hide them? It only hides them. Google says Search features cannot be turned off, so an extension removes the block from the page in your browser after it loads, and the AI Overview is still generated. ### Is it safe to install a Google AI Overview hider extension? We cannot vouch for any extension. Before you install, check the listing's developer details, its update date and its Privacy section, and once it is installed, limit its site access in Chrome's extension Details to the Google site you use. ### Why did my AI Overview extension stop working? Most likely Google changed its results page and the extension no longer finds the AI Overview. Check the listing for a recent update. The Firefox Hide Google AI Overviews listing, for example, said on 19 October 2026 that it had stopped working in late March and was waiting for an update to be reviewed; by 10 November 2026 that notice was gone and the listing showed a newer version. ### Can I hide AI Overviews in Chrome without an extension? Yes. Choose the Web filter after each search, or add a custom search engine in Chrome's settings that opens the Web view and make it your default. That shortcut relies on an address parameter Google does not document, so it may stop working. ## Sources - [Google Search Help: AI Overviews in Google Search](https://support.google.com/websearch/answer/14901683?hl=en) - [Chrome Web Store Help: Install and manage extensions](https://support.google.com/chrome_webstore/answer/2664769?hl=en) - [Chrome for Developers: Fill out the privacy fields (Chrome Web Store)](https://developer.chrome.com/docs/webstore/cws-dashboard-privacy) - [Chrome Web Store listing: Google AI Overview Hider (checked 19 October 2026)](https://chromewebstore.google.com/detail/aaafgcgohpoahgdhhgkpodjfekjgbfln) - [Chrome Web Store listing: AI Overview Hider for Google (checked 19 October 2026)](https://chromewebstore.google.com/detail/foobohnghnhkmgpglaefdnbcjkenjpgi) - [Chrome Web Store listing: Hide Google AI Overview (checked 19 October 2026)](https://chromewebstore.google.com/detail/ckalpgoholmlfjncjahmnkmmihodgajc) - [Firefox Add-ons listing: Hide Google AI Overviews (checked 10 November 2026)](https://addons.mozilla.org/en-US/firefox/addon/hide-google-ai-overviews/) - [Google Chrome Help: Set default search engine and site search shortcuts](https://support.google.com/chrome/answer/95426?hl=en) - [udm14.com (third-party shortcut site by Tedium)](https://udm14.com/) --- # Guide: Jasper AI and Surfer SEO: how they differ, how they work together, and whether you need both Canonical: https://whatisgeo.app/jasper-ai-surfer-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-30 > Jasper mainly generates on-brand marketing copy; Surfer SEO mainly scores a draft against the pages that already rank. When we checked on 30 October 2026, neither vendor documented a live integration between them. One produces words. The other compares those words with what already ranks. Everything this page says about either product comes from that vendor's own pages, which we read on 30 October 2026. ## What Jasper does and what Surfer SEO does Jasper's platform page has the heading "Jasper is AI built to execute marketing end to end". It says that by combining agents, content pipelines, shared governance and human orchestration in one workspace, Jasper "helps teams execute marketing with AI agents, at scale". Its content pipelines run in stages from planning through creation to activation. A central feature is Brand Voice, which Jasper says helps teams "prevent brand voice violations before they happen". In short, Jasper's main job is producing a lot of marketing copy that stays on-brand. Surfer's main job is search optimisation. Its help centre describes the Content Editor as "a tool that helps you write and optimize your articles to achieve higher search engine rankings". Surfer says the Content Editor builds its guidelines from top-ranking pages and its own algorithms, and gives live feedback through a Content Score, an SEO Score and an AI Search Score. The split between writer and grader is a simplification, though. Surfer also documents Surfer AI, which it says "helps you create high-quality, SEO-friendly articles faster" and which you can start from the Content Editor. Jasper also has SEO pages of its own. Today each tool overlaps with the other. What still differs is the starting point: Jasper starts from your brand, Surfer from the search results. ## Jasper vs Surfer SEO at a glance | | Jasper | Surfer SEO | |---|---|---| | Main job (vendor's description) | Executing marketing work with AI agents at scale | Writing and optimising articles for search rankings | | Starting point | Your brand voice, style guide and product knowledge | A keyword and the pages that already rank for it | | Main output | Draft copy for campaigns, SEO and other channels | Guidelines, scores and, through Surfer AI, full articles | | AI search features (vendor's description) | GEO & AI Optimization: measuring brand visibility in AI answer engines | AI Visibility: monitoring brand mentions in AI answers | | Documented link to the other tool | None found (30 October 2026) | None found (30 October 2026) | We have left prices out on purpose. Plans change often, and only the vendors' own pricing pages are reliable. Check them on the day you buy. For a wider look at the category, see our [AI SEO tools guide](/ai-seo-tools/). ## How the two connect, according to each vendor Neither vendor currently lists the other as an integration. When we checked on 30 October 2026, Jasper's integrations page listed tools including Semrush, Webflow, Google Docs and Slack, and Surfer was not among them. The integrations section of Surfer's help centre covers its API, MCP, Google Docs, WordPress and Contentful, and does not mention Jasper. We can't tell from this whether something still works inside either app for older accounts. What we can say is that neither vendor's current integrations list names the other. If a blog post or review says the two connect, check its date and confirm against the vendors' own pages. You can still combine the tools by hand. Draft in Jasper, paste the draft into Surfer's Content Editor, edit until you are happy, then publish. That is two separate products and two subscriptions. ## Do you need both? Most people don't, at least to begin with. Work out which problem you actually have. - **You need lots of on-brand copy across channels and a team.** Jasper's brand voice and governance features are aimed at that problem. - **You need to know whether a specific page holds up against what ranks.** Surfer's Content Editor is aimed at that problem. - **You need search-led articles drafted for you.** Surfer AI does this inside Surfer, so you may not need a separate writing tool. Paying for both only makes sense if you have the first two problems at once. One example is a content team that writes a lot in a set brand voice and also wants every page checked against search results before it goes live. Even then, with no documented integration, you are combining the tools yourself. That is our recommendation, not a requirement from either vendor. ## What neither tool can promise Neither tool guarantees rankings or citations in AI answers. Jasper's GEO page has the headline "Get cited by AI", but that is marketing copy, and the page does not promise that any AI engine will cite you. Google's documentation for AI Overviews and AI Mode says there are "no additional requirements" to appear and no special optimisation is needed. A page has to be indexed and eligible to show with a snippet. Google also says that meeting its requirements does not guarantee your content will be crawled, indexed or shown. A high Surfer score means your page looks similar to pages that already rank. A polished Jasper draft means the copy reads well and matches your voice. Google documents neither as a ranking signal. Use them to help your judgement. For the AI search side, see [how to write content for AI search](/how-to-write-content-for-ai-search/) and the [principles section of our main guide](/guide/#principles). ## What Google says about AI-assisted content Google's guidance on generative AI content says AI can be useful for research and for structuring original content. It also warns that using AI to "generate many pages without adding value for users may violate" its spam policy on scaled content abuse. That policy covers unoriginal content made mainly to manipulate rankings that offers "little to no value to users, no matter how it's created". In practice, using Jasper or Surfer AI won't get you penalised just for using them. The risk comes from publishing a lot of low-value pages, whichever tool made them. We go into this in more depth in [is AI content good for SEO](/is-ai-content-good-for-seo/). If you are worried about AI detectors, our page on [Surfer's AI content detector](/surfer-seo-ai-content-detector/) explains what its score can and can't tell you. ## How to test them on your own pages Both vendors offer free trials or demos, so try them on your own pages before you commit. This is the method we recommend: 1. **Pick a few similar pages.** Choose pages of the same type and search intent, so the comparison is fair. 2. **Decide what you are measuring before you start.** Good options are time spent drafting, how many edits a draft needs before a person is happy with it, and how many factual errors you find. 3. **Change one thing per page.** Write one page in Jasper alone, one with Surfer's Content Editor alone, and one using both by hand. Keep notes. 4. **Check everything yourself.** Look at facts, tone and whether the page answers the reader's question. A score can't do that for you. 5. **Be careful about what you conclude from rankings.** Rankings change for many reasons, so don't give a tool the credit for a change you can't trace back to it. You are testing a workflow. Neither tool can sell you a result. ## From the same studio These are made by the studio that publishes this guide. - [ScriptGrain](https://scriptgrain.com/): If you are choosing a writing tool mainly to keep drafts in your own voice, ScriptGrain is our tool built for that job. - [visibilitytoai](https://visibilitytoai.com/): If what you want to know is whether AI assistants actually name and cite your company, VisibilityToAI measures that across ChatGPT, Claude, Gemini and Perplexity. ## Questions people ask ### Is Jasper or Surfer SEO better for SEO? For checking one page against search results, Surfer is the closer fit, because its Content Editor builds its guidelines from top-ranking pages. Jasper is built around producing on-brand marketing content. Neither one guarantees rankings. ### Does Jasper still integrate with Surfer SEO? When we checked on 30 October 2026, neither company's current integrations list named the other, so we could not find a current integration documented by either vendor. If you rely on an older connection, confirm with the vendors directly. ### Can I use Jasper and Surfer together, and do I pay for both? Yes. You can draft in one tool and paste the text into the other by hand. They are separate products sold on separate plans, so check each vendor's own pricing page before you buy. ### Will content written in Jasper and scored in Surfer rank? Nobody can promise that. Google says meeting its requirements does not guarantee a page will be indexed or shown, and its spam policies judge content on its value to users, no matter how it was created. ### Does Surfer SEO write content on its own? Yes. Surfer documents a feature called Surfer AI that generates full articles, which you can start from the Content Editor. Check Surfer's help centre to see what your plan includes. ## Sources - [Jasper: Platform overview (vendor page, read 30 October 2026)](https://www.jasper.ai/platform) - [Jasper: Brand Voice (vendor page, read 30 October 2026)](https://www.jasper.ai/brand-voice) - [Jasper: GEO & AI Optimization (vendor page, read 30 October 2026)](https://www.jasper.ai/geo) - [Jasper: Integrations (vendor page, read 30 October 2026)](https://www.jasper.ai/integrations) - [Surfer Help Center: What is Surfer AI?](https://docs.surferseo.com/en/articles/7869670-surfer-ai) - [Surfer Help Center: Content Editor overview](https://docs.surferseo.com/en/articles/5700347-content-editor-overview) - [Surfer Help Center: Integrations, MCP & API (read 30 October 2026)](https://docs.surferseo.com/en/collections/3932030-integrations-mcp-and-api) - [Surfer Help Center: AI Visibility](https://docs.surferseo.com/en/collections/16906691-ai-visibility) - [Google Search Central: Guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central: Spam policies for Google web search (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) --- # Guide: Jasper AI SEO: what Jasper does for SEO, what it cannot do, and what Google says Canonical: https://whatisgeo.app/jasper-ai-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-11-10 > Jasper is an AI marketing platform that drafts and optimises content and, on its Business plan, tracks AI visibility. Google judges content on quality, so no tool can guarantee rankings or citations. Jasper helps with the writing side of SEO, and on its Business plan it also reports on how AI assistants mention your brand. Whether any of that becomes rankings or citations depends on the pages you publish, and Google judges those the same way whoever or whatever wrote them. This page keeps three kinds of claim apart: what Jasper says about its own features (vendor claims, read on 31 October 2026), what Google documents, and our own recommendations. ## What "Jasper AI SEO" actually covers Jasper sells itself as an AI platform for marketing teams. The core is content creation in a workspace it calls Canvas, with your brand voice and company knowledge applied to what it writes. SEO is one of the things you use that platform for. In practice the search term covers three things: - **Drafting** content aimed at search. - **Agents** that research, score and restructure content for search engines and AI answers. - **Reporting** on how AI assistants describe and cite your brand, through a feature Jasper calls the GEO Hub. Jasper does not describe a rank tracker of its own. More on that below. ## What Jasper documents about its SEO features Everything in this section is Jasper's own description, taken from its SEO and AEO page, its GEO page, its help centre and its pricing page. We have not tested these features, so treat them as claims to check. - **SEO and AEO agents.** Jasper lists specialised agents for "the SEO/AEO content pipeline", including Competitor Audit, Gap Finder, Entity Mapper, AI Readiness Score, Fact Density Audit, Schema Markup (which generates JSON-LD), Query Planner, FAQ Generator and Pillar Article. - **GEO Agent.** Jasper says it "identifies the questions people actually ask AI engines, then creates answer-first content to match", and adds FAQs, citations and schema to existing content. - **Semrush integration.** The help centre says connecting Semrush gives the GEO Agent and Studio Agents "real keyword, competitive, and SERP data". Both agents work without it, but then they rely on "general knowledge". The integration needs a Semrush Business plan (Advanced on newer Semrush pricing) plus a purchased API unit pack, and Semrush API units are separate from Jasper credits. Semrush Pro and Guru plans cannot use it. - **Jasper Grid.** Jasper describes this as a way to "generate or refresh pages in bulk". - **GEO Hub.** The help centre says it shows how your brand appears in AI responses across ChatGPT, Gemini, Claude and Perplexity. It reports a composite 0 to 100 visibility score, brand presence rate, citation rate, competitive share of voice and sentiment, then suggests pages to fix or create. It is available on Business plans, runs on credits and refreshes daily, weekly or monthly. - **Google AI Overviews.** Jasper's SEO and AEO page says you can "track your presence and attribution in AI Overviews alongside traditional search metrics". The GEO Hub help article we read lists only ChatGPT, Gemini, Claude and Perplexity as providers, so ask Jasper how it collects AI Overviews data before relying on it. Scores such as AI Readiness Score are Jasper's own measures. None of the Google documentation we read refers to them. ## What Google says about AI-written content Google's position is less dramatic than the forums make it sound. - **The tool is not the test.** Google's February 2023 guidance says it focuses "on the quality of content, rather than how content is produced", and that "appropriate use of AI or automation is not against our guidelines". Using automation, including AI, to generate content "with the primary purpose of manipulating ranking in search results" breaks its spam policies. - **Scaled content abuse.** Google's spam policies define this as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users", "no matter how it's created". The first example given is "using generative AI tools or other similar tools to generate many pages without adding value for users". The same policies also cover attempts to manipulate generative AI responses in Google Search. - **Fact-check everything.** Google's guidance on generative AI content says it is "critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing", and that this applies to titles, meta descriptions, structured data and image alt text too. - **People-first content and E-E-A-T.** Google's helpful content guidance lists "Are you using extensive automation to produce content on many topics?" as a warning sign. It says trust is the most important part of E-E-A-T (experience, expertise, authoritativeness and trustworthiness), and that its quality raters treat "using generative AI to produce large amounts of text without manual oversight or curation" as little to no effort. So the useful question is whether a page is made for people or made to game a result. Bulk features such as Grid are where that question matters most. For the wider argument, see [is AI content good for SEO](/is-ai-content-good-for-seo/). ## Where Jasper helps and where it stops | Job | What Jasper documents | What to bear in mind | |---|---|---| | Drafting on-brand content | The core of the product | You still supply the facts and first-hand experience | | Keyword research | Through the Semrush integration | Needs a separate Semrush Business plan and API units | | Rank tracking | No rank tracker of its own on the pages we read; Semrush tools such as Organic Positions can be added to Studio Agents | Google Search Console or an SEO suite remains the usual source | | Schema markup | An agent that generates JSON-LD | Google says structured data must match the visible text; validate it | | Mentions in ChatGPT, Gemini, Claude and Perplexity | GEO Hub, Business plan, paid in credits | Jasper's own method, on queries you choose | | Google AI Overviews | Claimed on the SEO and AEO page | Google counts AI Overviews and AI Mode traffic in Search Console's Performance report, under the "Web" search type | | Rankings or citations | Not guaranteed | Nobody can guarantee them | If you are weighing Jasper against other tools, [AI SEO tools](/ai-seo-tools/) explains what each type of tool can and cannot tell you. ## A safe way to use Jasper for SEO content This checklist is our recommendation. It follows from what Google documents; Jasper does not set it. 1. **Start with a real reader need.** Decide who the page is for and what they are trying to do before you open the tool. 2. **Treat the draft as raw material.** Jasper gives you a first pass; you give it substance. 3. **Add what the tool cannot.** Your own experience, examples from your work, and detail a reader could not get from a generic summary. 4. **Check every claim.** Include the title, meta description and any generated schema, as Google asks. 5. **Cap bulk work at what you can review by hand.** If you use Grid to generate or refresh many pages, each one still has to add value. Google calls it scaled content abuse when many pages are generated mainly to manipulate rankings without helping users. 6. **Do not chase Jasper's scores for their own sake.** A higher readiness or fact density score is only worth having if the facts are true and sourced. 7. **Be open about how the page was made.** Google suggests AI disclosures where readers would reasonably ask "How was this created?", and its guidance on helpful, people-first content says fabricating creator profiles, "such as by using AI-generated headshots, made-up names, or false credentials", is a form of deception. Notice what is missing: any trick. The safe route is the boring one. Make something useful, then use the tool to make it faster. ## Does Jasper help you appear in AI answers? It can help you write and structure pages. It cannot put you in an answer. Google says a page is eligible to appear as a supporting link in AI Overviews or AI Mode if it is indexed and eligible to be shown in Search with a snippet, and that "there are no additional technical requirements". It also says specific optimisation is not required for these features. So there is no special AI markup a writing tool can add to make a page eligible. The foundations are the same as for ordinary search. Measuring whether ChatGPT, Gemini, Claude or Perplexity mention you is a separate job. The GEO Hub is one way to do it inside Jasper, if you are on the Business plan. Whatever you use, a visibility score reflects a sample of queries chosen by you or the vendor, so check a few answers yourself. [How to measure AI visibility](/how-to-measure-ai-visibility/) and our guide to an [AI visibility checker](/ai-visibility-checker/) cover the methods. ## What to check before you pay On 31 October 2026 Jasper's pricing page listed two plans. **Pro** costs $69 per seat per month billed monthly, or $59 per month billed yearly with a 12 month commitment. It includes one seat and has a 7-day free trial. **Business** has custom pricing, starts with a 12 month commitment, and adds agents for GEO, Jasper Grid and API access. Jasper says usage-based features such as the GEO Hub and agents run on credits alongside seat pricing on Business. Prices change, so check the page on the day you buy. 1. **Match features to plans.** The GEO Hub and Grid are listed under Business, not Pro. 2. **Price the credits.** Ask what your report frequency and page count would cost on Jasper's credits rate card. 3. **Price Semrush too** if you want live keyword data, since the integration needs Semrush's Business plan and API units. 4. **Ask how the GEO Hub picks its queries,** and compare a few of its results with answers you check by hand. 5. **Decide how you will measure SEO results.** Search Console reports Google traffic, including AI Overviews and AI Mode. 6. **Treat Reddit threads and agency reviews as anecdote.** One person's ranking jump or detector result is not evidence about yours. The thinking behind how we judge tools like this is set out in [our principles](/guide/#principles). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want an independent check of whether ChatGPT, Claude, Gemini and Perplexity name and cite your company, alongside or instead of a vendor's own score. - [ScriptGrain](https://scriptgrain.com/): If you are choosing an AI writing tool for SEO content, it drafts in your own measured voice for you to edit and fact-check. ## Questions people ask ### Is Jasper AI good for SEO? It is built for drafting and optimising content, and Jasper documents SEO agents and a Semrush integration for research. Whether pages rank depends on their quality and usefulness, which Google judges however they were written. ### Does Google penalise content written with Jasper? Not for being AI-written. Google says appropriate use of AI is not against its guidelines. Generating many pages mainly to manipulate rankings, without adding value for users, is scaled content abuse whatever tool made them. ### Can Jasper do keyword research or track rankings? Jasper says its agents can use Semrush keyword, competitor and SERP data if you connect a Semrush Business plan with API units. The Jasper pages we read describe no rank tracker of its own, so use Search Console or an SEO suite for that. ### Does Jasper help with AI Overviews or ChatGPT citations? Jasper's GEO Hub, on the Business plan, reports how often ChatGPT, Gemini, Claude and Perplexity mention and cite your brand, using Jasper's own method. Google says AI Overviews eligibility needs only normal indexing and snippet eligibility, so no tool can guarantee inclusion. ### Do I need Surfer SEO as well as Jasper? Not necessarily. Jasper documents its own optimisation agents and a Semrush integration, so check whether those cover what you need before paying for a second tool. ## Sources - [Jasper: Plans and pricing (vendor page, read 31 October 2026)](https://www.jasper.ai/pricing) - [Jasper: Optimize for SEO, AEO and more (vendor page, read 31 October 2026)](https://www.jasper.ai/solutions/seo-aeo-geo) - [Jasper: GEO and AI search optimisation (vendor page, read 31 October 2026)](https://www.jasper.ai/geo) - [Jasper help centre: GEO Hub](https://help.jasper.ai/hc/en-us/articles/51579082265883-GEO-Hub) - [Jasper help centre: Integrations: Semrush](https://help.jasper.ai/hc/en-us/articles/53731968390555-Integrations-Semrush) - [Google Search Central: Guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central: Spam policies for Google web search (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: Creating helpful, reliable, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) --- # Guide: LLM SEO: what it means and what to actually do Canonical: https://whatisgeo.app/llm-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > LLM SEO is another name for GEO: making your content findable, quotable and accurately described in AI answers. It builds on ordinary SEO, and nobody can guarantee inclusion. In practice, LLM SEO means two jobs: letting AI search tools reach and understand your pages, and writing pages clear enough to quote. Most of that is SEO you may already be doing. The rest of this page sorts the usual tactics by how much evidence stands behind each one. ## What LLM SEO means (and the other names for it) LLM SEO is the practice of making it more likely that AI tools find your pages, use them as sources and describe your business correctly. It is an extension of ordinary SEO, with no hidden layer of the web behind it. You will see it under several labels: LLM optimisation, SEO for LLMs, answer engine optimisation (AEO) and generative engine optimisation (GEO). They all describe the same work. The fuller definition is in [what GEO is](/guide/#what-is-geo), and how it relates to traditional search is covered in [GEO vs SEO](/guide/#geo-vs-seo). Most "LLM SEO" guides hand you a list of tactics and treat them all as equally proven. This page sorts them into three kinds of claim instead: - **Platform-documented:** the company that runs the system has said so in its own documentation. - **Research finding:** someone tested it, under conditions that limit how far it applies. - **Sensible practice:** a recommendation. No platform requires it. ## Two different things people call LLM optimisation "Optimising for an LLM" can mean two separate goals, and they need different expectations. **Being retrieved and cited by live AI search.** Someone asks a question, the tool searches or fetches pages, and your page becomes one of the sources it uses or links to. You can influence this. Your pages need to be reachable by the right crawlers, indexed, and clear enough to quote. **Being known from model training.** A model learns from a large body of text collected before it was released. You cannot submit your site to a training run, and you cannot see what went in. The platforms do let you opt out of training crawlers, but controlling access is a different thing from controlling what a model later says about you. So most of the practical work is on the retrieval side. On the training side, about the most you can do is be described clearly and consistently across the web over time. Be sceptical of anyone who claims more control than that. ## What the platforms themselves document The platforms have published more about this than many guides admit. Everything in this section is a documented platform statement. **Google.** Google's "AI features and your website" page says that to be eligible as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to show in Google Search with a snippet, and that there are no additional technical requirements. It also says specific optimisation is not required for these features, while existing SEO fundamentals are still worthwhile. Its examples include allowing crawling in robots.txt and in your CDN or hosting setup, internal linking, and making sure important content is available as text. **OpenAI.** OpenAI documents separate crawlers. Sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers. Disallowing GPTBot signals that your content should not be used to train its foundation models. ChatGPT-User makes visits a user has asked for, and OpenAI says robots.txt rules may not apply to it. OpenAI also says each setting is independent of the others, so you can allow search and block training. **Perplexity.** PerplexityBot is the crawler that surfaces and links websites in Perplexity's search results, and Perplexity says it is not used to crawl content for AI foundation models. Perplexity-User handles user actions, and Perplexity says that because a user requested the fetch, it generally ignores robots.txt rules. **Anthropic.** ClaudeBot collects web content that may be used for model training. Claude-SearchBot indexes content to improve search results. Claude-User fetches pages when a person asks Claude something. Anthropic says its bots honour robots.txt, and that blocking Claude-SearchBot or Claude-User may reduce your visibility in search and user-directed answers. **Google-Extended.** This is a robots.txt token that controls whether Google uses your content to train Gemini models and to ground Gemini answers. Google says it does not affect inclusion in Google Search and is not a ranking signal, so blocking it will not remove you from Search. **robots.txt itself.** RFC 9309, the standard for robots.txt, says its rules are not a form of access authorisation. Crawlers are asked to honour them, and as the OpenAI and Perplexity notes above show, fetchers acting for a user may not. For setting these rules in practice, see [crawler access](/guide/#crawler-access). **llms.txt.** This is a proposal published at llmstxt.org: a Markdown file, usually at your site root, that gives AI agents a short summary and links to your key pages. The proposal itself does not say any major AI platform reads these files, and none of the platform documents above mention it. ## What the research says, and its limits The best-known study is Aggarwal et al., "GEO: Generative Engine Optimization", accepted at KDD 2024. The authors built a benchmark of queries (GEO-bench) and tested nine ways of rewriting source content, then measured how visible each source was in generated answers. What they reported: - Adding quotations, adding statistics and citing sources did best among the methods tested. - Keyword stuffing gave little or no improvement. The authors suggest that tactics that work in traditional search may not carry over. - Which methods worked best varied by subject area. The conditions matter. Their main generative engine was one they built themselves: gpt-3.5-turbo writing answers from the top five Google results. They also ran a smaller test on Perplexity.ai. They say the methods may need to change as these engines change, and that they did not measure effects on search rankings. These are results from a controlled experiment, not a measurement of ChatGPT, Google or Claude as they work today. So treat the paper as a set of ideas worth testing. Quotations, figures and named sources also help human readers, so there is little to lose by using them where they are genuine. Do not read it as a promise that any particular edit will increase how often you are cited. ## What to do: sensible practice Everything in this section is a recommendation. Where a platform or the research backs a step, that is noted. | Step | Kind of claim | |---|---| | Get pages indexed and eligible for snippets | Platform-documented (Google) | | Decide which search and training crawlers to allow | Platform-documented (OpenAI, Perplexity, Anthropic, Google) | | Keep important content in text, not only in images or scripts | Platform-documented (Google) | | Answer the question near the top of the page | Sensible practice | | Back claims with named sources, quotes and real figures | Research finding, under the paper's conditions | | Describe your business the same way everywhere | Sensible practice | | Keep pages current and specific | Sensible practice | 1. **Get the basics right first.** Check that your important pages are indexed and nothing blocks them from showing with a snippet. 2. **Set crawler access on purpose.** Search crawlers and training crawlers are separate, so you can allow OAI-SearchBot, PerplexityBot and Claude-SearchBot while making your own decision about GPTBot, ClaudeBot and Google-Extended. Check your CDN or firewall too, since Google notes they can block crawling. 3. **Answer early.** Put the direct answer near the top, then the detail. A page that states its point plainly is easier to quote accurately. 4. **Source your claims.** Say where each figure comes from. It helps readers, and it matches what the GEO paper tested. 5. **Describe yourself consistently.** Use the same name, the same description of what you do and the same key facts on your site and your profiles. When those differ, an AI tool has conflicting material to work from. 6. **Keep pages specific.** Generic copy gives a system nothing distinctive to quote. Most of this overlaps with good SEO, and that is intended. The wider approach is in the [principles](/guide/#principles) section, and if you are starting from scratch, the [30-day plan](/guide/#getting-started) puts these steps in order. ## What not to bother with, and claims to distrust **Do not build a plan around llms.txt.** It is cheap to add, but nothing in the platform documentation says the major AI tools use it, so do not expect it to change anything. **Do not believe guaranteed citations.** No one outside these companies controls what an AI answer includes, and the platforms themselves promise nothing. **Distrust unsourced numbers.** If a guide says a tactic delivers a specific uplift, ask what was tested, on which system and under what conditions. If the figure comes from a vendor selling the service, treat it as that vendor's claim. **Skip keyword stuffing.** It is old SEO advice, and the GEO paper found it did little in generated answers. **Be wary of tricks.** A tactic that only works by gaming one system tends to stop working when that system changes. Pages that are genuinely useful hold up better. ## How to tell whether it is working Measure what you can and be honest about the rest. Check that your pages are indexed. Look in your server logs for the documented crawlers by name. Ask AI tools the questions your customers ask, and note whether you are cited and whether you are described correctly. AI answers vary from one run to the next, so a single check proves little. Repeat the same set of questions over time and look for patterns across many questions rather than any single result. The method is set out under [measuring progress](/guide/#measurement). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once the basics are in place, the next question is whether ChatGPT, Claude, Gemini and Perplexity actually name and cite you, and this measures that. ## Questions people ask ### What is LLM SEO? It is the work of making your pages findable, quotable and accurately described in answers from AI tools such as ChatGPT, Perplexity, Claude and Google's AI features. It is another name for GEO, and it builds on ordinary SEO. ### Is LLM SEO different from GEO or AEO? No. LLM SEO, LLM optimisation, AEO and GEO are different labels for the same practice. Each label puts the stress somewhere slightly different, but the work is the same. ### Is SEO dead because of AI? No. Google's own documentation says pages appear as supporting links in AI Overviews and AI Mode by being indexed and eligible to show with a snippet, and that existing SEO fundamentals are still worthwhile. ### How do I do LLM SEO for my business? Make sure your pages are indexed, decide which AI crawlers to allow in robots.txt, put clear answers near the top of your pages, back your claims with named sources and describe your business the same way everywhere. Then check AI answers regularly over time. ### Can I guarantee ChatGPT will mention my site? No. You can let OpenAI's search crawler reach your pages and make them worth quoting, but no one outside OpenAI controls what an answer includes, and OpenAI offers no guarantee. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://developers.openai.com/api/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic help centre: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Google: Google's common crawlers (Google-Extended)](https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [llmstxt.org: The /llms.txt file proposal](https://llmstxt.org/) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: Perplexity SEO: how to get cited in Perplexity, and what is actually documented Canonical: https://whatisgeo.app/perplexity-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Perplexity searches the web and cites its sources with numbered links. The one step it documents is letting PerplexityBot crawl your site. The rest is research or practice, and nobody can guarantee a citation. Perplexity works like a search engine that writes an answer, so the work starts where ordinary SEO starts: can its crawler reach your page, and does the page clearly answer the question someone asked? Perplexity does not publish ranking factors beyond that, so this page puts every claim into one of three groups. **Documented** means Perplexity or a standards body publishes it. **Researched** means a peer-reviewed paper tested it, under conditions we state. **Recommended** is our own advice, labelled as such. ## What people mean by Perplexity SEO The phrase covers two jobs. Most people mean getting their pages cited in Perplexity's answers, and that is what most of this page covers. Some mean using Perplexity as a research tool for SEO work, which has its own short section near the end. Many of the top results for this search are agency pages and listicles. They present ranking factors as fact and rarely say where those factors came from. Where a point below has no source behind it, we say so. ## How Perplexity finds and cites pages Perplexity's help centre describes three steps: it interprets the question, searches the web, and then compiles the most relevant information into an answer. It says every answer includes numbered citations that link to the original sources, so readers can check the information. Follow-up questions keep the context of the conversation. The public documentation stops there. Neither the help centre nor the crawler page says how Perplexity chooses which sources to cite, or in what order. You can make a page eligible to be found and read, but you cannot decide whether it gets cited. For how AI answers are put together more generally, see [how AI answers happen](/guide/#how-it-works). ## What Perplexity documents about its crawlers Perplexity's crawler page lists two user agents: | User agent | What Perplexity says it does | robots.txt | | --- | --- | --- | | PerplexityBot | Surfaces and links websites in Perplexity's search results. Perplexity says it is not used to crawl content for AI foundation models. | Perplexity recommends allowing it in robots.txt so your site can appear in search results | | Perplexity-User | Visits a page when a user's question needs it, and may link to that page in the answer. Perplexity says it is not used for web crawling or to collect content for training. | Because a user requested the fetch, Perplexity says it "generally ignores robots.txt rules" | The full user-agent strings, as Perplexity publishes them: ``` Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; PerplexityBot/1.0; +https://perplexity.ai/perplexitybot) Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Perplexity-User/1.0; +https://perplexity.ai/perplexity-user) ``` In robots.txt, you refer to them by the short names `PerplexityBot` and `Perplexity-User`. Perplexity also publishes each agent's IP addresses as a JSON file, at perplexity.com/perplexitybot.json and perplexity.com/perplexity-user.json, and says these lists are updated regularly. For sites behind a web application firewall (WAF), it recommends allowing requests that match both the user agent and those IP ranges, and the page gives setup steps for Cloudflare and AWS WAF. Checking both is also how you tell a real Perplexity request from one that only borrows the name. ## Letting Perplexity in: robots.txt, PerplexityBot and Perplexity-User The one step Perplexity documents is to allow PerplexityBot. Robots.txt is defined in RFC 9309, and three points from that standard matter here. - **Rules are requests.** The RFC says crawlers "are requested to honor" the rules, and that the rules "are not a form of access authorization". Anything private needs a login or similar protection. A Disallow line will not keep it private. - **A named group replaces the catch-all.** A crawler follows the group whose user-agent line matches its name, and the match ignores upper and lower case. It falls back to the `User-agent: *` group only if no group names it. So a `PerplexityBot` group replaces your catch-all rules for that bot. The two are not combined. - **The most specific path wins.** Within a group, the Allow or Disallow rule with the longest matching path applies. If no rule matches, the path is allowed. For example, a site that blocks crawlers by default but wants to appear in Perplexity's search results could use: ``` User-agent: * Disallow: / User-agent: PerplexityBot Allow: / ``` The example only shows how the groups interact. Whether to allow the bot is your decision. Then work through these checks: 1. **Read your live robots.txt** at yourdomain.com/robots.txt. Look for a `PerplexityBot` group. If there is none, look for a catch-all `Disallow: /`, which then applies to it. Ready-made "block AI bots" templates often include PerplexityBot. 2. **Check your firewall or CDN.** Bot protection can turn away a crawler that robots.txt allows. Use Perplexity's published IP lists, as its documentation describes. 3. **Do not rely on robots.txt for Perplexity-User.** Perplexity says this agent generally ignores robots.txt, so restricting it takes firewall or server rules. 4. **Watch your server logs** for both user agents after any change. Our [AI crawler access guide](/guide/#crawler-access) covers the other platforms' bots. ## What research says about getting cited The most widely quoted study is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024. It is one of the few studies that tested Perplexity directly, so its conditions are worth knowing. **What they tested.** Nine ways of rewriting a page's text, including adding statistics, adding quotations, citing sources, simplifying the language, improving fluency and keyword stuffing. **How they tested it.** Most results come from the authors' own generative engine. For each query, it took the top five Google results and had a language model write an answer from them. The queries came from GEO-bench, the authors' benchmark of 10,000 queries across 25 domains, and some of those queries were taken from Perplexity's Discover section. Visibility was measured with the authors' own metrics, partly scored by GPT-3.5. They then ran the same methods on Perplexity.ai itself. **What they found.** In their setup, citing sources, adding quotations and adding statistics worked best. The authors report visibility gains of up to 40% on their own engine and up to 37% on Perplexity.ai. Keyword stuffing gave little or no improvement, and results varied by subject area. **What it does not show.** - "Visibility" is the authors' own measure of how much, and how prominently, an answer drew on a source it had already retrieved. It does not measure traffic or sales, and it says nothing about how to get retrieved in the first place. - Perplexity has changed since 2024. The study tells you how things worked under its conditions at the time, and the results may not hold today. - The gains came from adding genuine citations, quotations and figures. Making them up would be dishonest and would cost you readers' trust. ## Sensible practice (recommendations, not guarantees) Perplexity documents none of the following, and none of it comes with a promise. These are our recommendations. 1. **Decide on crawlers deliberately.** Choose whether to allow PerplexityBot yourself. Do not let a template make the choice for you. 2. **Keep key content in server-rendered HTML.** Text that only appears after scripts run may never be seen by some crawlers. 3. **Answer the question near the top.** A page that gives its answer early is easier for people and for software to use. 4. **Cite your own sources.** It helps readers, and it matches one of the methods that did well in the study above. 5. **Keep facts current and consistent.** Out-of-date prices, opening hours or product details give any summary something wrong to repeat. 6. **Earn credible mentions elsewhere.** Perplexity searches the wider web, so what reputable sites say about you can end up in an answer. This is a reasonable bet, but Perplexity does not name it as a factor. The reasoning behind these points is in our [GEO principles](/guide/#principles). ## How to check whether Perplexity cites you 1. **List your buyers' real questions**, in the words they use. 2. **Ask each question in Perplexity and open the numbered citations.** Record the date, whether your site was cited, whether your business was named without a citation, and who was cited instead. 3. **Repeat with the same list later.** Treat each answer as one data point. A pattern across many questions tells you more. 4. **Check your server logs** for PerplexityBot and Perplexity-User. A visit does not mean a citation. If PerplexityBot never visits, check robots.txt and your firewall first. To make this a regular routine, see [measuring progress](/guide/#measurement). ## Using Perplexity as a tool for SEO work This is the second meaning of the phrase. Perplexity's help centre lists features that suit research: numbered citations on every answer, follow-up questions that keep context, Pro Search for more in-depth answers with a choice of AI model (free users get a limited number each day), a Research mode that runs many searches and writes a report, and, on the Pro plan, tools for building reports and spreadsheets. In SEO work, people use these to research a topic before writing, to see which sources Perplexity cites for the questions they want to rank for, and to summarise competitors' pages. Features change, so check the help centre for current details. One caveat, which is our practice advice: open the cited source and confirm the claim is actually there before using it in a brief or a client report. Any tool that summarises pages can get a summary wrong. ## What nobody can promise Nobody outside Perplexity can promise you a citation, whether that is an agency, a consultant or a tool. Perplexity does not publish how it chooses sources, so anyone quoting a precise formula is guessing. You can control three things: whether PerplexityBot can reach you, how clearly your pages answer questions, and whether your facts are accurate. If someone quotes a citation rate or a traffic figure, ask who measured it and how. If the figure comes from an agency, it is that agency's number. It is not a fact about Perplexity. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you have checked a few questions by hand, it tracks whether Perplexity and other assistants name and cite your company for your buyers' questions over time. ## Questions people ask ### Is Perplexity good for SEO? It depends which meaning you have in mind. As a place to be cited, it is one more way readers can find you, with no guarantee attached. As a research tool, it is useful if you open its cited sources and check them. ### How do I get my website cited in Perplexity? Nobody can guarantee it. Perplexity's documented advice is to allow PerplexityBot in robots.txt and let its published IP ranges through your firewall. Beyond that, clear, current pages that answer the question directly are our recommendation. ### Does Perplexity follow robots.txt? Perplexity recommends allowing PerplexityBot in robots.txt if you want to appear in its search results. It says Perplexity-User, which fetches pages when a user asks a question, generally ignores robots.txt rules because a user requested the fetch. Under RFC 9309, robots.txt rules are requests and give no access control, so use a firewall for anything that must be blocked. ### What is the difference between PerplexityBot and Perplexity-User? PerplexityBot surfaces and links websites in Perplexity's search results. Perplexity-User visits a page when a user's question needs it and may link that page in the answer. Perplexity says neither is used to collect content for training AI foundation models. ### Is SEO dead now with AI? No. Perplexity describes searching the web and citing the pages it uses, so pages that can be crawled and clearly answer a question still matter. What has changed is that a reader may meet you inside a cited answer before they visit your site. ## Sources - [Perplexity: Perplexity Crawlers](https://docs.perplexity.ai/docs/resources/perplexity-crawlers) - [Perplexity Help Center: How does Perplexity work?](https://www.perplexity.ai/help-center/en/articles/10352895-how-does-perplexity-work) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: SE Ranking's AI visibility tracker: what it does, what its numbers mean, and how to check them Canonical: https://whatisgeo.app/se-ranking-ai-visibility-tracker/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-22 > SE Ranking tracks AI answers in two places: the AI Results Tracker inside its SEO platform and SE Visible, a standalone dashboard. Both report on prompts you choose, so treat the figures as samples and check them. SE Ranking sells AI tracking in two forms. The first is prompt tracking inside its main SEO platform, called the AI Results Tracker and grouped under what SE Ranking calls its AI Search Toolkit. The second is SE Visible, a standalone dashboard at visible.seranking.com. Below is what each covers, what its numbers mean and how to check them. Everything here about the products comes from SE Ranking's own product, help, API and pricing pages, read on 22 October 2026. Where we give advice, we say it is ours. ## What SE Ranking's AI visibility tracking is | | AI Results Tracker | SE Visible | |---|---|---| | Where it lives | Inside SE Ranking projects, next to rank tracking and audits | A separate product, described as "Powered by SE Ranking" | | Engines SE Ranking names | AI Overviews, AI Mode, ChatGPT and Gemini (product page and API docs) | ChatGPT, Gemini, AI Mode, AI Overviews and Perplexity, with "Claude will be added soon" | | How it is sold | Prompt limits in the Core and Growth plans, plus an AI Search add-on | Its own plans, or 200 prompts of access through the AI Search add-on | In short, the AI Results Tracker suits someone who already uses SE Ranking for SEO. SE Visible is a brand-level dashboard you can buy on its own. ## Which AI platforms it covers SE Ranking's own pages disagree on this, so check it during a trial. The AI visibility product page names AI Overviews, AI Mode, ChatGPT and Gemini, and the AI Results Tracker API documentation lists the same four. The pricing page shows "5 LLMs to track" under both the Core and Growth plans, yet the AI Search add-on on the same page names only "AI Overviews, AI Mode, ChatGPT". SE Visible's pages list five engines, Perplexity included, and say every plan gets all five. None of these pages listed Claude or Microsoft Copilot as tracked when we read them. If your buyers use an assistant that is missing, no dashboard figure will tell you about it. ## What it tracks and what it reports The two products use different metrics with similar-sounding names, so read the definitions before you compare them. ### AI Results Tracker SE Ranking's help article on the Rankings tab defines: - **Mention**: whether your brand was named in the AI answer, even without a link. - **Link**: whether your page was linked in the answer, and in which position. - **Top 3 Presence**: the share of AI answers where your mention or link appears in the top three positions. - **Source Presence**: the share of answers that contain at least one source link. - **Organic-AI Overlap** (AI Overviews only): how many unique URLs an AI Overview shares with the top 20 organic results for the same keyword. The percentage always divides by 20, and a day with no AI Overview shows as empty rather than zero. Each result links to a cached copy of the AI answer, and arrows show the change since the previous check. The API documentation says the full answer text is kept for 30 days and the presence data for one year. ### SE Visible SE Visible's own explainer defines: - **Visibility**: how often answers mention you, weighted by where the mention appears. First place scores 1.0, lower positions score less, and answers that leave you out score zero. - **Mentions**: whether your brand appears, with linked and unlinked mentions counted separately. - **Share of voice**: your share of mentions compared with the competitors you track. - **Average position**: where you usually appear in the answers that mention you. - **Net sentiment**: the share of positive mentions minus the share of negative ones, on a scale from -100 to +100. - **Citations**: tracked separately in a Sources view, where a source's coverage is the share of analysed answers that cite it. ## How it collects the data SE Visible documents more than most tools do. Its explainer says it checks every tracked prompt on every selected engine daily by default, with weekly checks available on some plans. It collects answers "through a real browser session in each engine's user interface" instead of through an API, and stores each response unedited. Each project tracks one brand in one location, with its own prompt language. You choose up to five topics, and industry prompt suggestions arrive pre-selected. For the AI Results Tracker, the product page says it "sends queries to AI platforms and captures the responses as a real user would see them". The help article says your prompt limit is shared across every project in the account. Neither product's documentation says how many times each prompt is run per check. That matters, because one answer is one draw. Ask SE Ranking before you treat a daily figure as stable. ## What it costs These are the prices SE Ranking's pricing pages showed us in US dollars on 22 October 2026. The main pricing page showed euro prices when we read it from another location, so you may see a different currency and amount. | Plan or add-on | Monthly billing | Billed annually (per month) | AI tracking included | |---|---|---|---| | SE Ranking Core | $129 | $103.20 | 100 prompts tracked daily, 5 LLMs | | SE Ranking Growth | $279 | $223.20 | 250 prompts tracked daily, 5 LLMs | | AI Search add-on | from $89 | from $71.20 | AI Results Tracker, AI competitor research, SE Visible access (200 prompts) | | SE Visible Basic / Core / Plus | $99 / $189 / $355 | "Save 20%", per the page | 200 / 450 / 1,000 prompts | SE Ranking offers a 14-day free trial of the main platform that includes 20 AI prompts a day. SE Visible advertises a 10-day free trial. Enterprise pricing is not published. Prices change, so check both pricing pages on the day you buy. ## What the numbers can and cannot tell you This is our reading of the definitions above. - **They describe your prompt list, not the market.** Add prompts where you are strong and every rate rises. None of it measures how many real people asked those questions. - **Weighted scores are not percentages of answers.** An SE Visible visibility score of 0.5 does not mean you appear in half the answers, because position changes the weight. - **Share of voice depends on who you track.** Add or remove a competitor and your share moves, even if no AI answer changed. - **A single day's arrow is weak evidence.** AI answers can change between runs, and the documentation we read does not say how many runs make up one check. Look at trends over weeks. - **Coverage differs by product.** A figure from the AI Results Tracker and one from SE Visible may cover different engines, so do not add them together. Our guide to [how to measure AI visibility](/how-to-measure-ai-visibility/) explains why repeated samples beat single checks. ## How to check the data yourself This is our recommendation, not a vendor requirement. It works with or without a paid tool. 1. Pick ten tracked prompts: some where the tool says you appear and some where it says you do not. 2. Open the cached copy of each answer in the tool and note whether you are named, linked and described correctly. Do this within 30 days, while the answer text is still kept. 3. Run the same prompts yourself on each engine, in a logged-out or temporary session, from the country the project tracks. Repeat each prompt at least three times. 4. Record in a spreadsheet whether you were named, whether you were cited and which other sources appeared. 5. Compare. If your hand checks and the dashboard broadly agree, you can lean on the trend. If they often disagree, ask SE Ranking whether location, timing or run counts explain the gap. Our [AI visibility checker](/ai-visibility-checker/) guide sets out a free way to run these spot checks, and [how to track AI Overviews](/how-to-track-ai-overviews/) covers Google's own surfaces. ## Questions to ask any tracker before paying These are our recommendations, not a formal standard. They apply to SE Ranking and to every option in our [generative engine optimisation tools](/generative-engine-optimization-tools/) guide: 1. Which engines does my plan cover, and where is that written down? 2. Are the prompts mine, the vendor's suggestions, or both? 3. How often is each prompt run, and how many times per check? 4. How exactly are mention, citation, position and share figures calculated? 5. Can I see or export the raw answers behind every number? 6. Which locations and languages can I track? A tool earns its cost when you need many prompts, several engines and a daily record you could not keep by hand. For a dozen prompts checked monthly, a spreadsheet is enough. ## What to do with the results Start with the gaps: prompts where you are never named, or named but never cited. Those are your to-do list. For Google's AI features, Google Search Central says "there are no additional requirements to appear in AI Overviews or AI Mode" beyond normal Search eligibility. It also says sites shown in them are reported in Search Console's Performance report, within the "Web" search type. So Search Console remains the place to see Google traffic, and crawlable, indexed, useful pages remain the starting point. The [measurement section](/guide/#measurement) of our main guide shows how tracker data, referral traffic and Search Console fit together. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want a second, independent reading to compare against any tracker, it checks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions. ## Questions people ask ### What is SE Ranking's AI visibility tracker, and is SE Visible the same thing? No. The AI Results Tracker is part of SE Ranking's main SEO platform and is sold through plan prompt limits and an AI Search add-on. SE Visible is a separate dashboard, "Powered by SE Ranking", with its own plans and its own metrics. ### Which AI platforms does SE Ranking track? SE Ranking's product page and API docs name AI Overviews, AI Mode, ChatGPT and Gemini for the AI Results Tracker. SE Visible lists those plus Perplexity, with Claude "coming soon". Its pages disagree in places, so confirm coverage during a trial. ### How much does SE Ranking's AI visibility tracking cost? On 22 October 2026 the pricing page showed Core at $129 a month (100 daily prompts) and Growth at $279 (250 daily prompts), or 20% less billed annually. The AI Search add-on started at $89 a month, and SE Visible plans ran from $99 to $355 a month. Prices and currency vary, so check on the day you buy. ### How accurate is any AI visibility tracker? A tracker reports on the prompts you chose, on the days they ran, and AI answers can change between runs. Neither SE Ranking product documents how many times each prompt runs per check. Check a sample against the stored answers and your own repeated runs before you rely on a number. ### Can I track AI visibility without paying for a tool? Yes, on a small scale. Run a fixed list of prompts by hand several times each, record whether you are named and cited, and repeat monthly. For Google, Search Console counts traffic from AI Overviews and AI Mode within the Web search type of its Performance report. ## Sources - [SE Ranking: Subscription pricing (read 22 October 2026)](https://seranking.com/pricing.html) - [SE Ranking: AI Search Toolkit / AI visibility tracker product page](https://seranking.com/ai-visibility-tracker.html) - [SE Ranking Help Center: How to use AI Results Tracker: Rankings](https://help.seranking.com/hc/en-us/articles/16335399186460-How-to-use-AI-Results-Tracker-Rankings) - [SE Ranking API documentation: AI Results Tracker](https://seranking.com/api/project/ai-result-tracker/) - [SE Visible: product and pricing page](https://visible.seranking.com/) - [SE Visible: Inside SE Visible: what you see and where the data comes from (29 September 2026)](https://visible.seranking.com/blog/how-se-visible-works/) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) --- # Guide: SEO AI agents: what they are, what they can do, and where a human still has to decide Canonical: https://whatisgeo.app/seo-ai-agents/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-23 > An SEO AI agent is software that carries out multi-step SEO tasks, such as audits, keyword clustering or drafting, with little prompting. Judge one by what it is allowed to change and the checks you keep. There is no standard definition, so the useful question is what a given agent is allowed to do on its own. Reading and reporting carries little risk. Publishing or editing live pages without a person checking first is where Google's spam and AI content policies come in. ## What an SEO AI agent is, and how vendors use the term In practice, an agent takes a goal, breaks it into steps, uses tools such as a crawler, a keyword database or your CMS to carry them out, and keeps going without a new prompt at each step. A chatbot answers you. An agent works through a task. The line between the two is blurry, and each vendor draws it where its product sits. Three vendor pages show the range. These are their words, and each comes from a company with an agent product to sell: | Vendor | How its page describes SEO agents | |---|---| | Ahrefs (blog, May 2026) | "An AI SEO agent is software that actually does the SEO work, as opposed to just describing it." It adds that agents do not need full autonomy: "Most good ones have human approval steps built in." | | Gumloop (SEO automation page) | "AI agents that automate every part of your SEO workflows", built "for keyword research, competitor analysis, audits, content briefs, and more". | | Nightwatch (SEO AI Agent page) | "Your 24/7 SEO AI Agent that never sleeps", to "Automate your SEO research, technical audits, SEO tasks, site changes, and rank tracking". | One definition is about doing the work, one is about automating whole workflows, and one includes making site changes. Those are very different levels of trust to hand over, which is why the label alone tells you little. For the broader set of AI products aimed at SEO, see [AI SEO tools](/ai-seo-tools/) and [what is AI SEO](/what-is-ai-seo/). ## What agents are used for in SEO According to the vendor pages above, agents are used for: - **Keyword research and clustering** (Ahrefs, Gumloop, Nightwatch) - **Technical audits** (all three) - **Content work**: optimisation and scoring (Ahrefs), content briefs (Gumloop), content suggestions (Nightwatch) - **Internal linking** (Ahrefs) - **Rank tracking and reporting** (Ahrefs, Nightwatch) - **Competitor analysis or monitoring** (Gumloop, Nightwatch) That list describes what vendors say they aim at. It does not verify capabilities. A task on a product page tells you what the vendor is selling, and you still have to test how well the agent does it on your own site. ## Three levels of autonomy: analyse, suggest, act Sort every agent task by what it is allowed to touch. This framework is our recommendation, not a source claim. | Level | What the agent does | Example tasks | Main risk | |---|---|---|---| | 1. Analyse | Reads, crawls and reports. Nothing on your site changes. | Audits, rank reports, keyword clusters | A wrong conclusion, which you can check before acting | | 2. Suggest | Proposes changes that a person approves or rejects | Internal link suggestions, title rewrites, draft briefs | Reviews done in a hurry, so poor suggestions get waved through | | 3. Act | Edits or publishes without a person in between | Auto-published pages, bulk metadata changes, site changes | Mistakes reach live pages and readers at the speed the agent works | So ask any product which level each task runs at, and whether you can cap it. Start every agent at level 1. Move a task to level 2 once you have watched it be right. Treat level 3 as something that needs a specific reason and a way to undo it. ## What Google documents about automated and AI-generated content The risky end of that scale is where Google's published policies apply. **Scaled content abuse.** Google's spam policies define it as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users", producing content of little or no value to users "no matter how it's created". Its examples include "Using generative AI tools or other similar tools to generate many pages without adding value for users", scraping content to generate many pages, and stitching together content from different pages without adding value. An agent set to publish pages in bulk is the setup that policy describes if the pages add nothing for readers. Whether a given setup crosses the line is a judgement you have to make. **Using generative AI.** Google's guidance says generative AI "can be particularly useful when researching a topic, and to add structure to original content". It asks publishers to focus on "accuracy, quality, and relevance, especially when automatically generating the content", warns that outputs may contain inaccuracies (hallucinations), and says "It is critical to manually factcheck and review all AI-generated content for accuracy". That review applies to metadata too: titles, meta descriptions, structured data and image alt text. The page also suggests telling readers how automatically generated content was created, where that makes sense for them. **People-first content.** Google says its automated ranking systems are designed to prioritise helpful, reliable information for people. It says E-E-A-T is not a specific ranking factor in itself, and that trust is the most important part of it. Where automation is used to substantially generate content, it suggests asking yourself who made it, how it was made and why. For more on publishing AI-written pages, see [is AI content good for SEO](/is-ai-content-good-for-seo/). ## Do SEO AI agents work? What can and cannot be claimed This page cites no independent evidence of agent performance. We found no primary source or named study measuring ranking uplift, traffic gains or time saved from SEO agents, so we state none. The figures that do exist come from vendors. Nightwatch's page has a heading reading "Save 100+ Hours" but gives no method. The Ahrefs article gives timings and results from its own team's workflows, with no controlled comparison. These are the vendors' own results, drawn from conditions they chose, and they show what a vendor reports. They are not a measure of what an agent will do for you. A tool can still be useful without a proven ranking effect. An audit that finds real broken links has saved you time whether or not a position moves. Judge an agent by what you can check (it found these errors, it grouped these queries sensibly), not by promised outcomes. ## Checks to keep before an agent changes your site Our recommendation, not a claim from any source. Keep all five: 1. **Read-only by default.** Give the agent no write access to your CMS or code until you have decided a task needs it. 2. **Human approval before anything goes live.** No page is published or edited on the agent's say-so alone. Google's own guidance asks for manual review of AI-generated content, metadata included. 3. **A change log.** Record every edit with what changed and when, so any of it can be reversed. 4. **Fact checks.** Verify every claim, figure, quote and link the agent adds. Google warns that generative AI output can contain inaccuracies. 5. **Your own measurement.** Track outcomes with a method you control, such as Search Console data, your analytics or a fixed set of test questions, and do not rely only on the vendor's dashboard. ## Agents and AI search visibility: where they do and do not help Agents fit the level 1 groundwork: checking whether pages are crawlable and indexable, collecting what AI answers say about your business, and tracking which sources get cited. That is the same work described in [how to measure AI visibility](/how-to-measure-ai-visibility/). What no agent can do is guarantee a citation. Google says there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary", and that SEO best practices remain relevant there. An agent can help you do that work, but it cannot change how Google or an AI assistant picks its sources. The research is about content, not agents. The GEO paper by Aggarwal et al. (KDD 2024) tested ways of rewriting content on its own benchmark, GEO-bench, in the authors' own generative engine setup. It found that effects varied by domain. It did not study SEO agents and is not evidence that they improve visibility. The site's [principles](/guide/#principles) cover what the evidence does support. ## How to choose or build one: questions to ask Whether you are buying or building, ask: 1. Which tasks does it do, and at which level: analyse, suggest or act? 2. Can I make it read-only, and require approval before any change? 3. Is every change logged and reversible? 4. How does it check facts and links before it writes them? 5. What evidence exists beyond the vendor's own results? 6. Can I measure outcomes without relying on its reporting? If a vendor cannot answer the third and fifth clearly, treat that as your answer. For a wider comparison of AI visibility tools, see [generative engine optimisation tools](/generative-engine-optimization-tools/). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If the outcome you want to check is whether AI assistants name and cite your company, it is one way to measure that across ChatGPT, Claude, Gemini and Perplexity without relying on the agent's own reporting. ## Questions people ask ### Can AI agents do SEO? They can carry out parts of it. Vendors such as Ahrefs, Gumloop and Nightwatch describe agents for keyword research, audits, content work, internal linking and rank tracking, but those are vendor descriptions and no independent evidence of ranking results is cited here. ### Is there an AI that can do SEO? Several products are sold as SEO AI agents, and each covers a different set of tasks. None can guarantee rankings, so check which tasks a product does, what it is allowed to change, and who reviews its work. ### What is the difference between an SEO AI agent and an SEO tool or chatbot? A chatbot answers a prompt and a tool does one defined job when you run it. An agent works through several steps towards a goal and uses tools along the way. Vendors draw the line in different places, so ask what the product does. ### Is it safe to let an AI agent publish content or edit my site automatically? Our recommendation is no, not by default. Google says it is critical to manually fact-check and review all AI-generated content before publishing, and its spam policies cover generating many pages without adding value for users, however they are made. ### Will an SEO AI agent get my site cited in ChatGPT or AI Overviews? Nobody can promise that. Google says there are no special optimisations needed to appear in AI Overviews or AI Mode beyond SEO best practice, so an agent can help you audit and track, but it does not control whether you are cited. ## Sources - [Google Search Central: Spam policies (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: Google Search's guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central: Creating helpful, reliable, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Ahrefs blog: AI Agents for SEO (Louise Linehan, May 2026), vendor page](https://ahrefs.com/blog/ai-agents-for-seo/) - [Gumloop: SEO automation with AI agents, vendor page](https://www.gumloop.com/solutions/seo-automation) - [Nightwatch: SEO AI Agent (NightOwl), vendor page](https://nightwatch.io/seo-ai-agent/) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: SEO vs GEO vs AEO: what is the difference, and do you need all three? Canonical: https://whatisgeo.app/seo-vs-geo-vs-aeo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > SEO aims to rank in search results, AEO to be the direct answer, and GEO to be cited in AI-generated answers. They overlap heavily, no definition is official, and the same sound foundations serve all three. Short answer: treat SEO, AEO and GEO as three names for one job. That job is to be the page a system picks when someone asks a question, and to be easy to quote when it does. The labels help you describe where you want to appear. They are a poor reason to split your team, budget or plan three ways. ## The short answer: three labels, one overlapping job Someone asks a question. A system (a search engine, an answer box, a chatbot) goes looking for pages that answer it well. Your job is to be one of those pages, and to be crawlable, indexable and clear enough to be used. SEO, AEO and GEO each describe a slightly different corner of that job. They are informal labels, and most of the work underneath them is the same. That is our view, and the rest of this page shows where each part of it comes from. ## What each term means and where it came from **SEO** (search engine optimisation) is the established one: making pages easy for search engines to crawl, index and rank, so people find them in results. **AEO** (answer engine optimisation) points at a narrower prize: being the answer itself. Think featured snippets, answer boxes and voice responses, where the system gives one reply instead of a list of links. **GEO** (generative engine optimisation) points at AI-written answers, where a model composes a response and may cite or mention sources as it goes. It is the only one of the three with a named research paper behind it: Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024. More on what that paper does and does not show below. We found that the top-ranking pages for these terms use AEO and GEO inconsistently. That is our own observation, and we know of no official definition for either. Some treat AEO as a subset of GEO, some treat them as the same thing, and some tie them to different surfaces. For the longer treatment of the GEO side, see [what GEO is](/guide/#what-is-geo) and [GEO vs SEO](/guide/#geo-vs-seo). ## SEO vs AEO vs GEO at a glance | | SEO | AEO | GEO | |---|---|---|---| | Surface | Search results pages | Answer boxes, featured snippets, voice answers | AI-written answers that cite or mention sources | | Goal | Rank and earn the click | Be the single answer shown | Be cited or described accurately in the response | | What you can measure | Indexing, impressions, clicks, positions | Whether you hold the answer for chosen queries | Whether and how you are cited for a fixed set of prompts | Read the table as a map of emphasis. One page can serve all three columns at once, and good pages usually do. ## Where they overlap They overlap on most of the work. Google's guidance on its AI features (AI Overviews and AI Mode) is direct about this. To be eligible as a supporting link, a page must be indexed and eligible to be shown in Google Search with a snippet, and Google says there are no additional technical requirements. It also says specific optimisation is not required for these features, and that existing SEO fundamentals stay worthwhile: allowing crawling in robots.txt (and at your CDN or host), internal links, good page experience, important content available as text, and structured data that matches the visible page. So a crawlable, indexable, clearly written page with accurate information sits underneath all three labels. If that foundation is broken, no label fixes it. If it is sound, you have done most of the work for each. ## Where they genuinely differ They do differ, though by less than the sales pages suggest. - **The surface.** A ranked list of links, a single boxed answer and a written AI response are different places to appear, and you are represented differently in each. - **The unit of success.** In classic SEO you usually want the click. In AEO you may be the answer even if nobody clicks. In GEO you want to be cited or described correctly, which may or may not send anyone to your site. - **Measurement.** Rankings and clicks have mature tools. Citations in AI answers are harder to track, because answers vary between runs and across products. - **Crawler access.** AI companies run their own crawlers with separate purposes, so robots.txt choices matter in new ways (see below). ## What the platforms say, what research shows, and what is just practice Mixing these three kinds of claim is how a lot of confident nonsense gets published, so here they are kept apart. ### What the platforms document **Google.** For AI Overviews and AI Mode: indexed plus snippet-eligible, no extra technical requirements, SEO fundamentals still apply. Google says these features surface links to supporting websites, and that AI Overviews only appear when its systems judge them additive, so they often do not trigger. Google treats robots.txt rules for Googlebot as the control over crawling for Search, including its AI features. To limit what is shown from your pages, it points to `nosnippet`, `data-nosnippet`, `max-snippet` and `noindex`. Google-Extended is a separate control for AI training and grounding in some of Google's other systems. **OpenAI.** OpenAI documents separate user agents. OAI-SearchBot is for surfacing sites in ChatGPT's search features, and sites that block it will not be shown in ChatGPT search answers (though they can still appear as navigational links). GPTBot crawls content that may be used to train its foundation models. ChatGPT-User visits pages for certain user actions, and OpenAI notes that robots.txt rules may not apply to it because a user starts those visits. You can allow one and block another. **robots.txt itself.** RFC 9309 defines the rules crawlers are "requested to honor" and states plainly that they "are not a form of access authorization". It is a request that well-behaved crawlers follow. It does not lock anything. For the practical setup, see [crawler access](/guide/#crawler-access). Other AI companies publish their own crawler names, so check each operator's documentation before writing rules for it. ### What research shows The Aggarwal et al. paper set out GEO as a research problem. The authors built GEO-bench, a benchmark of 10,000 queries across domains, and tested ways of rewriting source pages to raise their visibility in generated answers. They report that adding citations, quotations from relevant sources and statistics helped visibility in their tests, and that what worked varied by domain. Know the conditions before you lean on it: most results come from the authors' own two-step generative engine setup, with visibility scored by metrics they defined (partly judged by GPT-3.5), plus a smaller test on Perplexity. It is a useful signal about what generative systems may favour. It is not a rulebook for today's ChatGPT, Gemini or AI Overviews, and it promises no result for your site. ### What is just practice The following is our recommendation. No platform documents it as a requirement: keep technical SEO healthy, put a clear direct answer near the top of each important page, keep facts accurate and sourced, decide deliberately which crawlers you allow, and describe your business the same way everywhere it appears. It is cheap, sensible, and comes with no guarantee. Our [principles](/guide/#principles) and [how AI answers happen](/guide/#how-it-works) go into more depth. ## Do you need all three? A sensible order of work You need the work. You do not need three programmes. This is the order we would follow: 1. **Fix the foundation.** Pages crawlable, indexed and eligible for snippets. This is SEO, and Google's documentation says it is also what makes a page eligible for its AI features. 2. **Write for the answer.** Put a clear, direct response near the top of each page that matters. This serves answer boxes, and in our view it makes a page easier for any system to quote. 3. **Keep facts accurate and sourced.** Dull, and probably the most durable item on this list. 4. **Set crawler rules on purpose.** Decide which search and AI crawlers you want, using each operator's documentation, then set robots.txt to match. 5. **Make your description consistent.** Your name, what you do and who it is for should read the same on your site, your profiles and your listings. None of these steps needs its own budget line per acronym. They are one body of work under three labels. ## How to tell whether any of it is working Start with what is well established. Google says visits from AI Overviews and AI Mode are counted in Search Console's Performance report, within the "Web" search type, alongside the rest of your search traffic. Indexing, impressions and clicks there tell you whether the foundation holds. Answer boxes and AI citations are rougher to measure. Pick a fixed set of questions your customers actually ask, check how you appear for them on a regular schedule, and record what you see, including how you are described. Treat each check as a sample, because answers vary from one run to the next. Our notes on [measuring progress](/guide/#measurement) cover this in more detail. Be wary of anyone who promises a result. Nobody can guarantee your site will be cited in an AI answer, and Google's own documentation describes eligibility, not placement. ## Questions The common questions on SEO, AEO and GEO are answered below. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once the foundations are in place, it checks whether ChatGPT, Claude, Gemini and Perplexity actually name and cite your company for the questions your buyers ask. ## Questions people ask ### Is AEO part of GEO, or the other way round? Neither has an official definition, so it depends on who is writing. Some top-ranking pages treat AEO as a subset of GEO, others reverse it, and some use the two interchangeably (that is our observation, not a documented fact). Pick a definition, say what it is, and move on. ### Is GEO just SEO with a new name? Partly. Google says the same foundational SEO best practices apply to AI Overviews and AI Mode, with no additional technical requirements. GEO is still a useful label because the thing you are trying to earn is a citation or mention in a written answer, which may or may not bring a click. ### Do I need separate strategies for SEO, AEO and GEO? In our view, no. One plan covers all three: healthy technical SEO, clear direct answers, accurate sourced facts, deliberate crawler rules and a consistent description of your business. ### What about AIO, LLMO and 'AI SEO'? Are they the same thing? They are more informal labels for the same overlapping job, and we have not found an official definition for any of them. Check what a given writer or supplier means before assuming two of them are different services. ### Can anyone guarantee my site will be cited in an AI answer? No. Google describes indexing and snippet eligibility as what makes a page eligible, and says AI Overviews often do not trigger at all. Good foundations improve your chances, and nobody can promise a citation. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full text (HTML, v3)](https://arxiv.org/html/2311.09735v3) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) --- # Guide: Surfer SEO AI: what its AI features do, what they cost, and what to check before you pay Canonical: https://whatisgeo.app/surfer-seo-ai/ Publisher: Adapt Progress Evolve Limited Updated: 2026-11-10 > Yes: Surfer is a paid content tool with AI writing and scoring, plus an AI Tracker for brand mentions in AI answers. Plans start at $49 a month billed yearly (checked 25 October 2026), with a 7-day trial. Surfer sells two different kinds of AI product under one brand, so each needs its own verdict. The writing and scoring tools help you produce and tune pages. The AI Tracker reports how often AI assistants mention you. Neither can guarantee rankings or citations, and Google's own documentation explains why. Throughout this page we keep three kinds of claim apart: what Surfer says about its own product, what Google documents, and what independent research found. Prices and plan details come from Surfer's pricing page and help centre, read on 25 October 2026. Surfer shows prices in a currency chosen for the visitor: read from Europe on 10 November 2026, the page showed the same figures in euros. ## Is Surfer SEO an AI tool? Yes. Surfer's pricing page pitches it to "teams that want to win AI search", and every plan it lists includes AI features. In Surfer's own words, these include: - **Content Editor and Content Score:** you write or paste a draft for a target keyword and get live guidelines and a score from 0 to 100. - **AI writing (Surfy):** an AI writing assistant inside the editor. Surfer also lists "1-click content optimization". - **AI Detector and Humanizer:** a classifier that estimates whether text is machine-written, and a rewriter. We cover both on separate pages: the [Surfer AI content detector](/surfer-seo-ai-content-detector/) and the [free AI humanizer](/surfer-seo-free-ai-humanizer/). - **AI Tracker:** monitoring of brand mentions in ChatGPT, Perplexity, Google AI Mode, Google AI Overviews and Gemini. It reports a Visibility Score, mention gaps, share of voice and sentiment. Calling it "an AI tool" is accurate but not very useful. What matters is which of these you would use and what it is supposed to change. ## The two things "Surfer AI" can mean Search the phrase and you land on two products sharing one name. The first is the content side: the editor, the score and the AI writing assistant. It helps you write, and its output is a page. The second is the AI Tracker. It does not write anything. It runs a set of prompts through AI assistants and reports whether your brand appears. Surfer also sells it on its own as a plan called AI Search Analytics. It is easy to blur the two. Praise for the editor tells you nothing about the tracker, and the reverse is also true. Judge each against the job you need done. ## What the AI writing and content features do, and what Google says about AI content **How Surfer says the score works.** According to Surfer's help centre, Content Score has two parts. The SEO Score looks at "keyword usage, topical coverage, structure, and alignment with top-performing pages for your target query". It gives most weight to terms that "appear consistently across top-performing pages". The AI Search Score is based on "Facts Coverage" and "Upfront Intent Alignment": whether you cover the facts AI answers tend to include, and whether you answer the main question early. Surfer notes that the AI Search Score does not yet take competitors into account. The FAQ on its pricing page also says the guidelines draw on "over 500 web and AI signals". That figure is Surfer's own, and Surfer does not publish the list. Two details from the same help article are worth knowing before a trial. First, Surfer advises aiming 10 to 20 points above competitors instead of chasing 100, and says over-optimising "can actually hurt your final content". Second, the score needs competitor pages from at least three different domains, so branded keywords and other narrow searches can return a score of 0. **What that means in practice.** A score built from top-ranking pages tells you how closely your draft resembles what already ranks. That makes a reasonable coverage checklist. It cannot tell you whether your page is accurate, original or more useful than the others, and that is what Google says it rewards. **What Google says.** Google's 2023 guidance is titled "Rewarding high-quality content, however it is produced" and says "appropriate use of AI or automation is not against our guidelines". Its current page on generative AI content adds that it is "critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing". Its spam policies define scaled content abuse as generating many pages "for the primary purpose of manipulating search rankings and not helping users". One of the examples they list is "using generative AI tools or other similar tools to generate many pages without adding value for users". So by Google's documented rules, it is fine to use Surfy to draft a page that you then check, correct and improve with your own knowledge. Using any AI writer to turn out large numbers of thin pages is the pattern Google names as spam. The tool does not decide which side you are on; your process does. For the wider question, see [is AI content good for SEO](/is-ai-content-good-for-seo/). ## What the AI visibility tracker does, and how to read its numbers Surfer says the AI Tracker shows "how Gemini, ChatGPT, Google AI Overviews, AI Mode, and Perplexity talk about your brand". You add the prompts you care about. The tool then reports mentions, a Visibility Score, share of voice against competitors, sentiment, and the sources the AI answers drew on. Read its numbers with care. The points below are our advice; they are not from Surfer's documentation: - **It measures a sample.** A mention rate covers only the prompts in your project. Choose different prompts and the number changes. - **AI answers vary.** The same question can get different answers from one run to the next, so one day's figure is a snapshot. - **Coverage depends on the plan.** On the pricing page we read, the Standard plan card lists 25 prompts in ChatGPT, refreshed weekly. Pro lists 50 prompts refreshed daily across all five engines. - **Tools disagree.** Two trackers with different prompts and schedules can report different scores for the same brand without either being broken. Before you trust a figure, ask: - Which prompts are tracked, and who chose them? - How often do they run, and on which engines? - Does the report show the actual answers and citations, or only a headline score? Our guide to [measuring AI visibility](/how-to-measure-ai-visibility/) goes through this in more detail. **Can content changes move those numbers?** There is some research. Aggarwal et al. (GEO, KDD 2024) built a benchmark of 10,000 queries called GEO-bench. They tested nine ways of rewriting source pages in a generative engine they set up themselves, which fetched the top five Google results for each query. Adding citations, quotations and statistics did best, raising the paper's own visibility measures by up to 40%. Keyword stuffing did poorly. They also ran tests on Perplexity. The results varied by subject area, and they came from a research setup, so they may not carry over to today's commercial assistants. The paper does not test Surfer or any other commercial tool. Google's position on its own AI features is simpler: "specific optimization isn't required for AI Overviews and AI Mode". A page must be indexed and eligible to show in Search with a snippet, and "there are no additional technical requirements". Normal SEO practice applies. ## Can you use Surfer for free, and what does it cost? **Free:** We found no free plan on Surfer's pricing page on 25 October 2026. Surfer does offer a 7-day trial with Pro-level access. Its help centre explains the terms: - You choose a plan and enter card details at checkout. - The trial "does not cancel automatically". If you don't cancel within 7 days, you are charged for the plan you chose. - The trial is for new accounts only. - If you cancel, your account expires and you lose any stored data. The help centre also mentions a free AI Visibility Report. **Cost:** These are the prices shown on Surfer's pricing page on 25 October 2026, in US dollars per month, billed yearly: | Plan | Per month, billed yearly | What Surfer lists | |---|---|---| | Discovery | $49 | 120 documents, AI SEO guidelines, Surfy and Humanizer; no AI tracking listed | | Standard | $99 | 360 documents; AI tracking of 25 prompts in ChatGPT, refreshed weekly | | Pro | $182 | Standard features plus AI tracking of 50 prompts, refreshed daily, across ChatGPT, Perplexity, AI Mode, AI Overviews and Gemini | | Peace of Mind | $299 | "Unlimited" documents (Surfer's help centre sets a fair usage limit of 500 a month), 100 prompts daily, API access | | AI Search Analytics | $82 ($95 billed monthly) | AI Tracker only: 50 prompts daily across the same five engines | | Enterprise | Custom | Custom limits | Yearly plans are paid upfront for the whole year. Surfer says paying yearly saves "up to 17%" compared with monthly billing, so monthly prices for the content plans are higher than the figures above. Surfer also says UK and EU customers have VAT added unless they give a valid VAT ID. Prices change, so check Surfer's pricing page on the day you decide. ## How to test it yourself before you pay Seven days goes quickly, so plan the test before you start the trial. These steps are our recommendation: 1. **Set a reminder to cancel** on day six, because the trial turns into a paid plan automatically. 2. **Pick one real page** you already want to improve, on a topic you know well. 3. **Run it through the Content Editor.** Note every suggestion, accept the ones you agree with and reject the rest. Write down which was which and why. 4. **Fact-check anything Surfy writes** before it goes on the page, as Google's guidance asks. 5. **Time the work** and compare it with how long the same edit would take without the tool. 6. **If you care about the tracker,** list ten questions a real customer would ask and add them as prompts. Then type a few into ChatGPT and Perplexity yourself, and compare what you see with what the tracker reports. 7. **Decide on what you found.** If the suggestions held up and the work went faster, that is a reason to pay. If you cannot tell, that is an answer too. You will not see ranking changes inside a week. Judge the trial on whether the suggestions were useful and accurate. **Is it legit?** Surfer is a real company with published plans, a help centre and documented cancellation terms. It says it has more than 150,000 customers, which is its own figure. Review sites carry plenty of user opinions. Give more weight to reviewers who say what they used it for and for how long. Be cautious of reviews that repeat Surfer's marketing wording or send you to it through affiliate links. A star rating tells you whether people were satisfied, not whether the tool fits your work. ## Who it suits, and who can skip it The content tools suit people who: - publish regularly on topics they know - want a structured check of their coverage against what already ranks - will edit and fact-check AI drafts themselves The tracker suits teams who want a regular view of brand mentions in AI answers and understand that it reports on a chosen sample of prompts. You can probably skip it if you publish rarely, if your content lacks something people actually want, or if you expect the tool to supply expertise. A score cannot fix those. If you only need tracking, the standalone plan or a different tracker may fit better. Our overview of [AI SEO tools](/ai-seo-tools/) sets out the options without ranking them. Whatever you choose, the method is the same. Decide what you are trying to change, test on a real page, and judge by what you measured. Our home guide covers [how to measure progress](/guide/#measurement). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want a second, separate read on whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions, this measures it. ## Questions people ask ### Is Surfer SEO an AI tool? Yes. Surfer's own pages list AI writing (Surfy), an AI-informed Content Score, an AI detector and humanizer, and an AI Tracker that monitors brand mentions in ChatGPT, Perplexity and Google's AI features. The writing tools and the tracker do different jobs, so judge them separately. ### Can I use Surfer SEO for free? We found no free plan on Surfer's pricing page on 25 October 2026. New accounts get a 7-day trial with Pro access, but you have to enter card details, and Surfer charges you for the plan you chose unless you cancel before the trial ends. ### How much does Surfer SEO cost per month? On 25 October 2026 Surfer's pricing page showed Discovery at $49, Standard at $99, Pro at $182 and Peace of Mind at $299 a month, billed yearly. The standalone AI Search Analytics plan was $82 a month billed yearly or $95 billed monthly. Monthly billing costs more, and UK and EU customers pay VAT on top. ### Is Surfer SEO legit, and does it work? It is a real product with published plans, a help centre and documented cancellation terms. Whether it works for you depends on your process, so test it on one real page during the trial and judge it on how accurate and useful its suggestions are. ### Does content written with Surfer AI rank, or get cited in ChatGPT? No tool can guarantee either. Google says appropriate use of AI is not against its guidelines, but you should fact-check AI content, and mass-producing pages with no added value counts as spam. For AI Overviews and AI Mode, Google says no special optimisation is required. ## Sources - [Surfer: Pricing & Plans (vendor page, read 25 October 2026)](https://surferseo.com/pricing/) - [Surfer help centre: How does the Surfer trial work? (read 25 October 2026)](https://docs.surferseo.com/en/articles/12944181-how-does-the-surfer-trial-work) - [Surfer help centre: Pricing FAQ (read 25 October 2026)](https://docs.surferseo.com/en/articles/7545943-pricing-faq) - [Surfer help centre: AI Search Analytics (read 25 October 2026)](https://docs.surferseo.com/en/articles/12944184-ai-search-analytics) - [Surfer help centre: Fair Usage Policy (read 25 October 2026)](https://docs.surferseo.com/en/articles/12944161-fair-usage-policy) - [Surfer help centre: Content Score in the Editor Explained (read 25 October 2026)](https://docs.surferseo.com/en/articles/5700365-content-score-in-the-editor-explained) - [Surfer: AI Tracker (vendor page, read 25 October 2026)](https://surferseo.com/ai-tracker/) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central: Guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central: Spam policies for Google web search (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024, arXiv 2311.09735)](https://arxiv.org/abs/2311.09735) --- # Guide: Surfer SEO's AI content detector: what it does, how far to trust it, and what Google says Canonical: https://whatisgeo.app/surfer-seo-ai-content-detector/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-19 > Surfer's AI content detector gives a probability score for how likely text is machine-written. Treat it as an estimate: Google's documented guidance judges content on quality and intent, however it was made. The score shows how closely a text resembles machine writing, as judged by Surfer's model. It cannot tell you who wrote the text, and Google does not say it measures anything like it. This page keeps three kinds of claim apart: what Surfer says about its own tool (vendor claims), what independent research and OpenAI's own experience show about AI detection, and what Google documents about AI-generated content. ## What Surfer's AI content detector does Everything in this section is Surfer's own description, taken from its AI detector page on 19 October 2026. - **Input and output.** You paste text and the tool highlights the passages it judges to be AI-written. Surfer says it "assigns an AI probability score": a higher score suggests AI content, and a lower score suggests a higher chance of human writing. - **Method.** Surfer says the detector uses machine learning trained on "large datasets of both human-written and AI-generated texts", plus statistical analysis "such as n-grams and word frequency distributions". It also looks at features such as sentence difficulty, repetition and predictable structure. - **Language.** Surfer says the detector supports English only and "has only been trained on content written in English". - **Free and paid.** The page offers a free detector, with "unlimited AI detection" as part of a paid Surfer account. We have not checked current prices, so this page gives none. - **Accuracy.** The page describes the tool as "known for its accuracy" and quotes a customer testimonial calling it "one of the most accurate" tools they tested. The page we read gives no measured accuracy rate and no test method, so there is nothing to check those claims against. Surfer's own FAQ is more cautious than its headline. It says detectors "work pretty well, but they are not perfect", that false positives happen, that detectors "can misinterpret content created by non-native English speakers", and that "human oversight is necessary". Surfer also sells a tool for the other side. Its separate AI Humanizer page says it rewrites AI text into "human-like writing that can pass most of AI detection tools", and the detector page has a section headed "Avoid AI content detectors". So the same company offers one tool to flag AI text and another to make AI text harder to flag. Keep that in mind when you decide how much weight to give the score. ## How AI detectors work, and why scores are estimates A detector only sees the finished words. It has no access to drafts, edit history or the person who wrote them. It compares the text's patterns (word choice, predictability, sentence structure) with patterns it learned from examples, then outputs a probability. So a high score means "this looks like the machine-written text the model learned from". It does not mean that a set share of the text was written by AI, and it proves nothing. Short texts, heavily edited drafts and plain, formulaic human writing are all harder to call. Surfer's own FAQ also says detectors can struggle with output from more advanced AI models and have to keep adapting as models change. ## What research says about detector accuracy and bias **Liang et al. (2023), "GPT detectors are biased against non-native English writers".** The authors tested seven widely used detectors, including Originality.AI, GPTZero, ZeroGPT, Sapling and Crossplag. Surfer was not one of them. They ran the detectors on 91 TOEFL essays by non-native English writers and 88 essays by US eighth-grade students. The detectors were near-perfect on the US essays, but they labelled more than half of the TOEFL essays as AI-generated, an average false positive rate of 61.22%. When the researchers used ChatGPT to make the TOEFL essays' word choices more varied, the false positive rate fell sharply. When they simplified the word choices in the US essays, the false positive rate rose. The authors link this to perplexity, a measure of how predictable a text is to a language model. Plainer, more predictable language looks more "AI-like" to these detectors. The authors also set out the limits of their work: it is a pilot study with small samples, most of the detectors tested were built on GPT-2, and it covers only perplexity-based and supervised detection methods. So it cannot tell you how Surfer's detector behaves today. It does show that a detector can wrongly flag careful human writing, and the authors caution against using detectors in evaluative or educational settings. **OpenAI's withdrawn classifier.** OpenAI released its own AI text classifier in January 2023. A note now at the top of the announcement says that as of 20 July 2023 the classifier "is no longer available due to its low rate of accuracy". The original post said that on OpenAI's own test set the classifier correctly flagged 26% of AI-written text and wrongly labelled human-written text as AI-written 9% of the time. It also said the classifier was "very unreliable" on texts under 1,000 characters and "should not be used as a primary decision-making tool". Those figures are for OpenAI's tool, not Surfer's. They show how hard the problem is, even for the company that builds the models. ## Does Google use AI detection to rank or penalise pages? None of the Google Search documentation we read mentions AI detection scores as a ranking factor. Here is what it does say: - **Quality, however content is made.** Google's February 2023 guidance says it focuses on "the quality of content, rather than how content is produced", and that "appropriate use of AI or automation is not against our guidelines". - **Manipulation is spam.** Using automation, including AI, to generate content "with the primary purpose of manipulating ranking in search results" breaks Google's spam policies. - **Scaled content abuse.** This spam policy covers many pages "generated for the primary purpose of manipulating search rankings and not helping users", usually unoriginal, low-value content, "no matter how it's created". One of Google's examples is using generative AI to produce many pages that add no value for users. - **Fact-checking is expected.** Google's guidance on generative AI content says it is "critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing". So the documented test is value and intent. Human-written pages churned out at scale to manipulate rankings fall under the same policy as AI-written ones. For the wider question, see [is AI content good for SEO](/is-ai-content-good-for-seo/). ## How to use a detector score sensibly Our recommendation is to treat the score as a reason to look harder, never as a verdict. 1. **Note the score, then put it aside.** It tells you where to look. It does not make the decision for you. 2. **Read the highlighted passages yourself.** Are they generic, padded or vague? That matters to readers whoever wrote them. 3. **Check every fact.** Detectors do not check accuracy, and Google tells you to. 4. **Add what only you can add:** first-hand detail, a clear position, examples from your own work and named sources. 5. **Decide on quality.** Publish when the page is accurate and useful, whatever the number says. 6. **Never judge a person on a score alone.** Given the research above and Surfer's own caveats, a flag on a freelancer's or student's work is a reason to talk to them and look at their drafts, and nothing more. If you are comparing tools more broadly, our overview of [AI SEO tools](/ai-seo-tools/) explains what each type of tool can and cannot tell you. ## What to do instead of chasing a "human" score Rewriting text until a classifier passes it, whether by hand or with a humaniser, improves the score and nothing else. Google's guidance asks for something different: accuracy, originality and value to the reader. A page with those qualities is in good shape whatever a detector says. A page without them gains nothing from a better score. The Liang study also found that simple rewording could push detector results either way, which makes the score a weak guide to quality. Put the time into the page itself. Answer the question directly, back up claims with sources and add experience a reader could not find elsewhere. [How to write content for AI search](/how-to-write-content-for-ai-search/) covers this in practical detail, and the [principles section of our guide](/guide/#principles) explains the thinking behind it. ## Detectors and AI visibility AI visibility means whether assistants such as ChatGPT, Gemini, Perplexity and Google's AI features mention and cite you. Google says you can apply the same foundational SEO best practices to its AI features as to Search overall. Nothing in the Google documentation we have read, and nothing on Surfer's page, says a detector score affects whether a page is cited in an AI answer. A low AI score is not a visibility strategy. The documented guidance points to clear, accurate, well-sourced pages that crawlers can reach. ## Questions people ask ### Is Surfer SEO's AI content detector accurate? Surfer describes it as known for its accuracy, but the page we read publishes no measured accuracy rate or test method. Surfer's own FAQ says detectors are not perfect, produce false positives and need human oversight. ### Does Google penalise AI-generated content? Not for being AI-generated. Google says appropriate use of AI is not against its guidelines, and its spam policy on scaled content abuse targets low-value pages made mainly to manipulate rankings, however they are created. ### Can an AI detector prove a text was written by AI? No. A detector only estimates how closely a text resembles machine writing. Liang et al. (2023) found that the detectors they tested wrongly flagged more than half of a set of essays by non-native English writers as AI-generated. ### Does passing an AI detector help a page rank or get cited in AI answers? None of the Google documentation we have read says a detector score affects ranking or AI features. Google says the same foundational SEO practices apply to its AI features as to Search overall. ### Should I use an AI humaniser to get past a detector? We do not recommend it. Rewriting for a classifier does nothing for accuracy or value, which is what Google's guidance asks for. Time spent fact-checking and adding first-hand detail does more for readers. ## Sources - [Surfer: Free AI Detector (vendor page, read 19 October 2026)](https://surferseo.com/ai-content-detector/) - [Surfer: Free AI Humanizer (vendor page, read 19 October 2026)](https://surferseo.com/ai-humanizer/) - [Liang et al., GPT detectors are biased against non-native English writers (arXiv 2304.02819, 2023)](https://arxiv.org/abs/2304.02819) - [OpenAI: New AI classifier for indicating AI-written text (withdrawal note, July 2023)](https://openai.com/index/new-ai-classifier-for-indicating-ai-written-text/) - [Google Search Central: Guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central: Spam policies for Google web search (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) --- # Guide: Surfer SEO's free AI humanizer: what it does, and does it help your SEO? Canonical: https://whatisgeo.app/surfer-seo-free-ai-humanizer/ Publisher: Adapt Progress Evolve Limited Updated: 2026-11-10 > Surfer's AI humanizer is free up to a word limit and rewrites text so AI detectors score it as human. Google judges pages on helpfulness and its spam policies, so humanising alone will not make weak content rank. Surfer's humanizer is free to try within a word limit, and Surfer markets it partly on getting past AI detectors. Nothing Google publishes treats a detector score as a ranking signal, so a humanised page succeeds or fails on the same things as any other page: whether it is accurate, original and useful. ## What Surfer's free AI humanizer is Surfer's own tool page says the AI Humanizer takes AI-generated text and rewrites it into "natural, human-like text". Surfer's FAQ says it "uses sentence and style patterns typical of human writers", and that its goal is to help AI text "pass simpler AI detectors" and read more easily. On cost, the same page says (checked 21 October 2026): | Who | Humanizer limit, per Surfer | |---|---| | Not registered | 500 words | | Registered, no paid plan | 1,000 words | | Paid subscriber | Varies by plan | A notice on the page says the tool supports English only for now, although the FAQ further down mentions multiple languages. If language matters to you, test it before relying on it. Surfer's marketing goes further than its FAQ. The page says the tool lets text "bypass AI detection tools like Turnitin, Copyleaks and GPTZero" and that natural, human-like text is "the kind of content readers, AI search engines and Google trust". Those are Surfer's claims about its own product, and we have not seen independent testing behind them. Surfer's FAQ is more cautious. It says a humanizer's ability to evade detectors depends on how sophisticated both tools are, and that "more advanced" detectors might still pick the text up. Surfer also offers an [AI content detector](/surfer-seo-ai-content-detector/). One tool scores text; the other rewrites it so it scores differently. ## What a humanizer changes, and what it cannot change A humanizer works on surface features: sentence length, word choice, rhythm and the phrasing habits detectors tend to flag. Surfer's FAQ mentions removing passive voice and "other common AI patterns". A stiff paragraph often comes back reading more loosely. What it cannot change: 1. Whether the claims are true. 2. Whether the page answers the question the searcher actually asked. 3. Whether the page adds anything the existing results do not already say. 4. Whether someone with real experience stands behind it. A rewritten thin article is still a thin article. A rewrite can also introduce errors of its own, so check a humanised draft again before you publish it. ## What Google documents about AI-generated content Google sets out its position in a Search Central blog post from February 2023 and in a documentation page on generative AI content. - **Appropriate use is allowed.** The blog post's FAQ says: "Appropriate use of AI or automation is not against our guidelines." - **Quality is the test, however content is made.** Google says its ranking systems aim to reward "original, high-quality content", and that AI gets no special treatment: "Using AI doesn't give content any special gains. It's just content." - **Manipulation is spam.** Using automation, including AI, to generate content "with the primary purpose of manipulating ranking in search results" breaks Google's spam policies. - **Review is expected.** The documentation page says "it is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing." Google's spam policies define scaled content abuse as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users", "no matter how it's created". One listed example is scraping content to generate many pages, "including through automated transformations like synonymizing, translating, or other obfuscation techniques", where little value is provided. That example is specifically about scraped content. Even so, it shows that Google does not count automated rewording as something that makes a page valuable. Our wider guide, [is AI content good for SEO](/is-ai-content-good-for-seo/), covers these documents in more depth. ## Do AI detector scores matter for search? None of the Google pages above mentions third-party AI detectors or detector scores. They talk about helpfulness, accuracy, originality and the spam policies. A detector score is a number from another company, measuring something Google has never said it measures, and nothing Google has documented supports treating it as a ranking lever. Detectors also have known limits. The clearest public evidence comes from OpenAI, which withdrew its own AI text classifier on 20 July 2023 "due to its low rate of accuracy". Before that, OpenAI reported that on its own English test set the classifier correctly flagged 26% of AI-written text and wrongly labelled human-written text as AI-written 9% of the time. It also said the classifier was "very unreliable" on texts under 1,000 characters. Those figures describe only that one retired tool. Still, they show why a detector score is at best a probability and never proof. Our page on [Surfer's AI content detector](/surfer-seo-ai-content-detector/) covers what these tools can and cannot tell you. Humanizer vendors that promise to beat "all detectors" are making marketing claims. A sales page is not a test result. ## Where humanising text can sensibly help This section is our recommendation. If an AI draft reads flat, a rewrite pass can help. Repeated sentence shapes, stiff phrasing and an over-smooth tone are real problems for readers, and fixing them makes a page better to read. It can also help when a client or editor simply wants copy that does not sound machine-made. Surfer makes this point itself: it says Google does not punish AI content, but that readers and clients "may expect to see human writing". Use a humanizer as a copy-editing aid, and judge its output the way you would judge any edit. Academic and workplace rules on AI use are a separate matter that this page does not cover, so check the rules that apply to you. ## A better test than "does it pass a detector?" Before you publish, ask these questions in place of checking a detector score. They draw on Google's people-first guidance, but the list itself is our recommendation: 1. Does this page answer the query better than what is already there? 2. Is every claim checkable, and has someone checked it? 3. Does it contain something only you could add: experience, data, a clear view? 4. Would a named person put their name to it? 5. Is it one useful page, or one of many near-identical pages made to catch searches? If the answer to all five is yes, the detector score matters very little for search. If any answer is no, a rewrite will not fix it. The [principles in our main guide](/guide/#principles) start from the same place: write for the reader first, then check the facts. ## How this fits with AI search visibility Detector scores and AI search visibility are separate questions. Google's page on AI features says the same foundational SEO best practices apply to AI Overviews and AI Mode, that there are "no additional technical requirements", and that "specific optimization isn't required". A page that is useful, crawlable and accurate is what gets considered for citation, and how human a paragraph scores is not part of any documented requirement. Nobody can guarantee inclusion in an AI answer. If AI search is your goal, read [how to write content for AI search](/how-to-write-content-for-ai-search/) next, and see [AI SEO tools](/ai-seo-tools/) for the wider set of tools and what each one actually measures. ## From the same studio These are made by the studio that publishes this guide. - [ScriptGrain](https://scriptgrain.com/): If you draft with AI and want copy that already sounds like you before any rewrite pass, ScriptGrain drafts in your own measured voice for you to fact-check and edit. ## Questions people ask ### Does Surfer SEO have a free AI humanizer? Yes, within limits. Surfer's tool page says non-registered users can humanise up to 500 words and registered users without a plan up to 1,000, with higher limits on paid plans (checked 21 October 2026). ### Does humanising AI text help it rank on Google? Nothing Google publishes says it does. Google's guidance judges content on whether it is helpful, original and accurate, and says using AI gives content no special gains either way. Humanising can make a draft read better, but it does not fix weak content. ### Can Google detect AI-written content? Google does not document using an AI detector, and its stated position does not depend on one. It says appropriate use of AI is not against its guidelines, and its spam systems target content made mainly to manipulate rankings, however it is produced. ### Is it safe to use an AI humanizer to bypass AI detectors? No tool can promise a bypass, and even Surfer's own FAQ says advanced detectors might still flag humanised text. Bypass claims from humanizer vendors are marketing, not evidence, and a detector score is not part of anything Google documents for Search. ### What should I do instead of using a humanizer? Edit for substance: fact-check every claim, answer the query properly, add first-hand detail and have a named person stand behind the page. If the prose still reads stiff after that, a humanizer can serve as a final polish, followed by another accuracy check. ## Sources - [Surfer: Free AI Humanizer (tool page and FAQ, checked 10 November 2026)](https://surferseo.com/ai-humanizer/) - [Google Search Central: Google Search's guidance on using generative AI content on your website](https://developers.google.com/search/docs/fundamentals/using-gen-ai-content) - [Google Search Central Blog: Google Search's guidance about AI-generated content (February 2023)](https://developers.google.com/search/blog/2023/02/google-search-and-ai-content) - [Google Search Central: Spam policies for Google web search (scaled content abuse)](https://developers.google.com/search/docs/essentials/spam-policies) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: New AI classifier for indicating AI-written text (withdrawn July 2023)](https://openai.com/index/new-ai-classifier-for-indicating-ai-written-text/) --- # Guide: What are Google AI Overviews? Canonical: https://whatisgeo.app/what-are-google-ai-overviews/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > AI Overviews are AI-generated summaries that Google shows on some searches, with links to supporting web pages. Google says no special markup is needed to be included, and no one can guarantee it. They do not appear on every search. Google says AI Overviews are shown only when its systems decide they add something to the classic results, so on many queries there is none. This page sticks to what Google has documented and labels each point as one of three things: Google's own documentation, a research finding, or our recommendation. ## What AI Overviews are, in plain terms **Google's documentation.** Google's Search Help calls an AI Overview "an AI-generated snapshot with key information and links to dig deeper". Search Central lists AI Overviews and AI Mode as the two AI features in Google Search. Both show links to supporting web pages alongside the generated text. Google also says plainly that "AI Overviews can and will make mistakes", and it tells searchers to check important information in more than one place and to click through to the linked sources. For a site owner, the links are what count. An AI Overview sits among the normal results, which are still there, and the supporting links are how your pages appear in it. ## When did Google AI Overviews start? **Google's announcements.** The dates below come from Google's own blog posts: | Date | What Google announced | |---|---| | 14 May 2024 (Google I/O) | AI Overviews "will begin rolling out to everyone in the U.S.", following an experiment in Search Labs | | 15 August 2024 | Expansion to six more countries: the United Kingdom, India, Japan, Indonesia, Mexico and Brazil, with local language support | | 28 October 2024 | Expansion to more than 100 countries | According to the May 2024 post, AI Overviews were tested in Search Labs before launch. Google's Search Help now lists the countries, territories and languages where they are available. The list is long and it changes, so check that page for the current position. ## How an AI Overview is produced **Google's documentation.** Search Central says AI Overviews and AI Mode "may use a 'query fan-out' technique", which means "issuing multiple related searches across subtopics and data sources" to build a response. While the response is being generated, Google's models find more supporting pages. Google says this lets it show a "wider and more diverse set of helpful links" than a classic web search. Google has not published a list of signals that decide which pages get cited in an AI Overview. Anyone who claims to know the formula is guessing. Our [guide to how AI answers happen](/guide/#how-it-works) walks through retrieval and citation across the main assistants. ## How AI Overviews differ from AI Mode and featured snippets **Google's documentation.** | | AI Overviews | AI Mode | Featured snippets | |---|---|---|---| | What it is | AI-generated snapshot with links | Conversational AI response with links to supporting sites | A box that shows a page's snippet first, reversing the normal result format | | When it shows | Only when Google's systems judge it adds to classic Search | For questions that need more exploration, reasoning or comparison | Selected automatically by Google | | Source of the text | Generated by AI from several sources | Generated by AI from several sources | Taken from one page | Google says AI Mode and AI Overviews "may use different models and techniques", so their answers and links will vary. On mobile, Search Help says you can move from an AI Overview into AI Mode by selecting **Show more** and asking a follow-up question. ## What decides whether your site appears as a link **Google's documentation.** Search Central puts it this way: "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet". It adds: "There are no additional technical requirements." Google also says specific optimisation "isn't required" and that the usual SEO fundamentals still apply. Those include allowing crawling in robots.txt and at your CDN or host, linking to pages internally, putting important content in text, and keeping structured data consistent with the visible page. Eligibility does not mean inclusion. A page that meets every condition may still never be cited, and nobody can promise otherwise. **Research finding.** Aggarwal et al. ("GEO: Generative Engine Optimization", accepted to KDD 2024) tested changes to content, such as adding citations, quotations and statistics, and found that these could make a source more visible in generated answers. Bear in mind the conditions. The tests ran on a generative engine the authors built themselves, which answered from the top five Google results, using a benchmark of 10,000 queries. They repeated the tests on Perplexity.ai. They did not test Google AI Overviews, and they found that results varied by subject area. Treat it as a useful pointer, not a rule for Google. **Our recommendation.** Make sure your important pages are crawlable, indexed and able to show a snippet. Then put your effort into clear, well-sourced content in plain text. Paid markup or special files will not help, because Google says none is needed. Our [principles](/guide/#principles) explain why, and the [crawler access](/guide/#crawler-access) section covers robots.txt and CDN blocks. ## Can you turn AI Overviews off or opt out? Searchers and publishers get different answers. **Searchers (Google's documentation).** Searchers cannot turn them off. Google's Search Help says AI Overviews are "a core Google Search feature" and that "Features cannot be turned off." What you can do is select the **Web** filter after searching, which "displays only text-based links without features like AI Overviews". Turning off "AI in Search" in Labs only opts you out of experiments, not AI Overviews themselves. **Publishers (Google's documentation).** Google says robots.txt rules for Googlebot are how you control crawling for Search, AI features included. To limit what is shown from your pages, the robots meta tag documentation lists these controls: - `nosnippet`: no text snippet in any form of search result, including AI Overviews and AI Mode. Google says it "will also prevent the content from being used as a direct input for AI Overviews and AI Mode." - `max-snippet:[number]`: caps the snippet length and "will also limit how much of the content may be used as a direct input for AI Overviews and AI Mode." `max-snippet:0` works the same as `nosnippet`. - `data-nosnippet`: an HTML attribute that keeps specific parts of a page out of snippets. - `noindex`: removes the page from Google Search altogether. The catch is that these controls work across all of Search. `nosnippet` also removes the text snippet in ordinary web results, and `noindex` removes the page from Google entirely. Google offers no setting that only takes you out of AI Overviews. **Google-Extended does not cover this.** Google's crawler documentation says the Google-Extended token governs whether your content may be used to train Gemini models and to ground answers in Gemini Apps and Vertex AI. It also says it "does not impact a site's inclusion in Google Search". Blocking it will not remove you from AI Overviews. If you add a preview control and still see your content in AI features, Google suggests checking the control with the URL Inspection tool and waiting for a recrawl, which it says can take from several days to several months. ## How to see whether your site shows up **Google's documentation.** Search Central says traffic from AI features is counted in Search Console's Performance report, under the "Web" search type, together with the rest of your search traffic. Google's page does not mention a separate AI Overviews report, so you cannot use these figures to isolate AI Overview clicks. Google also suggests tracking conversions and time on site in tools such as Google Analytics. **Our recommendation.** Search for the questions your customers actually ask and note whether an AI Overview appears and which pages it links to. Repeat the check over time, and read it alongside your Search Console trend. Our [measurement guide](/guide/#measurement) explains how to do this without reading too much into small samples. ## What to do next These are our recommendations, based on Google's documentation: 1. Check your key pages in Search Console's URL Inspection tool to confirm they are indexed and nothing blocks a snippet. 2. Make sure robots.txt, your CDN and your host let Googlebot crawl them. 3. Before adding `nosnippet`, `max-snippet` or `noindex`, decide whether you can accept the effect on your ordinary results as well. 4. Keep important content in text, and keep structured data consistent with what visitors see. 5. Use the Performance report as your baseline, and run your own spot checks on the questions that matter to you. For a fuller plan covering every AI assistant, see our [30-day plan](/guide/#getting-started). ## Questions people ask ### What are Google AI Overviews? They are AI-generated snapshots of key information that Google shows in some search results, with links to web pages for more detail. Google shows them only when its systems judge that generative AI adds something to the classic results. ### When did Google AI Overviews start, and when did they reach the UK? Google announced on 14 May 2024 that AI Overviews were rolling out to everyone in the US, after an experiment in Search Labs. On 15 August 2024 Google announced they were coming to six more countries, including the United Kingdom. ### Can I turn off Google AI Overviews? Searchers cannot. Google's Search Help says features cannot be turned off, but the Web filter shows only text-based links. Publishers can use nosnippet, max-snippet or noindex, though these also affect how the page appears in ordinary results. ### Do I need special markup or a file to appear in an AI Overview? No. Google says a page must be indexed and eligible to show with a snippet, and that there are no additional technical requirements. Meeting those conditions makes a page eligible but does not guarantee it will appear. ### How are AI Overviews different from AI Mode? An AI Overview appears within normal search results and only on some queries. AI Mode is a conversational experience for questions that need more exploration or comparison. Google says the two may use different models and techniques, so their answers and links can differ. ### Does blocking Google-Extended remove my site from AI Overviews? No. Google's crawler documentation says Google-Extended does not affect a site's inclusion in Google Search. It controls use in Gemini model training and in grounding for Gemini Apps and Vertex AI. Googlebot and the snippet controls are what apply to Search. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Help: AI Overviews in Google Search](https://support.google.com/websearch/answer/14901683) - [Google Search Central: Robots meta tag, data-nosnippet and X-Robots-Tag specifications](https://developers.google.com/search/docs/crawling-indexing/robots-meta-tag) - [Google crawling infrastructure: Google's common crawlers (Google-Extended)](https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers) - [Google Search Central: Featured snippets and your website](https://developers.google.com/search/docs/appearance/featured-snippets) - [Google blog, 14 May 2024: Generative AI in Search (Google I/O 2024)](https://blog.google/products-and-platforms/products/search/generative-ai-google-search-may-2024/) - [Google blog, 15 August 2024: New ways to connect to the web with AI Overviews](https://blog.google/products-and-platforms/products/search/new-ways-to-connect-to-the-web-with-ai-overviews/) - [Google blog, 28 October 2024: AI Overviews expanding to more than 100 countries](https://blog.google/products-and-platforms/products/search/ai-overviews-search-october-2024/) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: What is AI SEO, and what is it called? Canonical: https://whatisgeo.app/what-is-ai-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > "AI SEO" means two things: optimising a site to be found and cited in AI answers, and using AI tools to do SEO work. The first is also called GEO, AEO or AI search optimisation, and no name is an official standard. When someone says "AI SEO", first find out which job they mean, because the advice for each is different. For the visibility job, Google's own documentation says its AI features need no special optimisation beyond normal SEO. The other AI companies document crawlers you can allow or block. ## The two meanings of AI SEO The first meaning is about **visibility**. People ask ChatGPT, Perplexity, Claude, Gemini or Google's AI Overviews a question and get a written answer, often with links to sources. The aim is for your pages to be among the sources those answers draw on and cite, and for your business to be described accurately. The second meaning is about **workflow**. You use AI tools to help with SEO tasks: outlining, drafting, summarising research, grouping keywords. Here AI is the assistant, and the target can still be ordinary search results. The two overlap, because you can use AI tools while optimising for AI answers. They are still separate questions, and most confusion in articles and sales pitches comes from treating them as one. The rest of this page is mostly about the first meaning, with a section on the second near the end. ## What SEO for AI search is called Several labels are in circulation. Publishers use them in overlapping and sometimes contradictory ways: | Name | Where it comes from or how it is used | |---|---| | GEO (generative engine optimisation) | Coined in a research paper by Aggarwal et al., first posted in 2023 and accepted to KDD 2024 | | AEO (answer engine optimisation) | Older term for featured snippets and voice answers, now stretched to cover AI answers | | LLMO (large language model optimisation) | Used by some writers for influencing what models know, as opposed to what they retrieve | | AI search optimisation (also AISO, ASO) | Plain descriptive labels with no single owner | | AI SEO | Either the umbrella for all of the above, or the workflow meaning | Two examples show how much the definitions vary. Search Engine Land's guide "What is AI SEO?" treats AI SEO as the broader discipline, with GEO, AEO and LLMO as parts of it. A separate Search Engine Land research piece, "SEO, GEO, or ASO?", surveyed marketers on which terms they recognise and use, and found no settled answer. Those are one publication's framings, not industry rules. Our reading is that no standards body defines any of these terms. The platforms we checked do not use any of them in their own documentation either: Google's guidance for its AI features simply talks about SEO. The same advice gets filed under every label. For the longer explanation of the term this site uses, see [what GEO is](/guide/#what-is-geo), and for how it compares with the classic discipline, [GEO vs SEO](/guide/#geo-vs-seo). ## What platforms actually document This is the most solid ground, because it is written down by the companies running the systems. ### Google AI Overviews and AI Mode Google Search Central's page on AI features says: - **No special optimisation is required.** It says specific optimisation is not required for AI Overviews and AI Mode, and that the same foundational SEO best practices apply as for Google Search overall. - **Eligibility.** To be shown as a supporting link, a page must be indexed and eligible to appear in Google Search with a snippet. It states there are no additional technical requirements. - **Controls.** Robots.txt rules for Googlebot control crawling for Search, including its AI features. To limit what is shown from your pages, use `nosnippet`, `data-nosnippet`, `max-snippet` or `noindex`. Google-Extended is the separate control for limiting AI training and grounding in some of Google's other systems. - **Measurement.** Traffic from AI features is included in Search Console's Performance report, under the "Web" search type. It is not broken out separately there. ### OpenAI, Anthropic and Perplexity Each company documents separate user agents for different jobs: | Company | Search crawler | Training crawler | User-triggered fetches | |---|---|---|---| | OpenAI | OAI-SearchBot: sites opted out will not be shown in ChatGPT search answers | GPTBot | ChatGPT-User: robots.txt rules may not apply | | Anthropic | Claude-SearchBot | ClaudeBot | Claude-User: blocking it may reduce visibility for user-directed search | | Perplexity | PerplexityBot: not used for training foundation models | None documented | Perplexity-User: generally ignores robots.txt | OpenAI says its settings are independent, so you can allow search while blocking training. Anthropic says its bots honour robots.txt. The practical point is that crawler access is a documented lever you can check and set. Much of what gets sold as AI SEO has no documentation behind it. More in [crawler access](/guide/#crawler-access). None of these pages promises that following them gets you into an answer. They describe eligibility and controls, not outcomes. ## What the research says, and its limits The paper by Aggarwal et al. introduced the term GEO and a benchmark called GEO-bench: a set of user queries across several domains, with web sources for answering them. The authors tested ways of changing page content and measured how visible each source was in generative engine responses. In their abstract they report that their methods boosted visibility by up to 40% on that benchmark, and that how well a method worked varied by domain. Read it as a study with conditions. Those results come from the authors' benchmark and their own visibility measures. They do not show that any one tactic works on today's ChatGPT, Google or Perplexity, for your queries or your site. The paper itself stresses that results differ across domains. Treat it as a source of ideas to test. ## What stays the same as classic SEO For Google, the documentation answers this directly: the same fundamentals apply. Its list includes allowing crawling in robots.txt and in any CDN or hosting setup, internal links, good page experience, important content available as text, structured data that matches the visible text, and up-to-date Business Profile and Merchant Center information. Beyond Google, the rest of this section is our recommendation, not a platform requirement. A page that is crawlable, clear, accurate and useful to the person asking is well placed for any system that retrieves web pages. A new interface on top does not change what a good source looks like. ## What is different in practice These are sensible practices, labelled as such. No platform requires them. - **Check robots.txt deliberately.** The AI companies now separate search, training and user-triggered agents, so an old blanket rule may be blocking search crawlers you would want, or allowing training you would not. Decide each one on purpose. - **Write answer-first.** State the answer near the top of the page, then support it. Readers find it clearer, and a self-contained answer is easier for any system to quote accurately. - **Think in full questions.** Plan pages around the sentences a buyer would actually type or say to an assistant, as well as short keywords. The home guide's [principles](/guide/#principles) cover these in more depth. ## Using AI tools to do SEO work This is the second meaning. AI tools can help draft, restructure, summarise and organise material you would otherwise do by hand. They do not check their own facts or take responsibility for what you publish. You still need to verify claims, keep the writing sounding like you and make sure each page helps a real reader. Speed makes it easy to publish many weak pages quickly, which serves nobody. ## How to tell whether it is working For Google, Search Console counts traffic from AI features within the Web search type, without a separate breakdown. For other assistants, the most direct check is sensible practice rather than a documented method. Put the real questions your buyers ask to the assistants they use, and note whether you are mentioned, whether you are cited and whether the description is accurate. Repeat this over time, because answers vary from one run to the next. Keep a dated log rather than a single screenshot. See [measurement](/guide/#measurement) for the wider picture. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want to check whether ChatGPT, Claude, Gemini and Perplexity actually name and cite your business for buyers' questions, this measures it. ## Questions people ask ### What is AI SEO called now? Several names are in use for the visibility meaning, including GEO, AEO, LLMO, AISO and ASO, and publishers define them in overlapping ways. None is an official standard, and Google's own documentation simply talks about SEO for its AI features. ### Is AI SEO the same as GEO? It depends who is writing. Search Engine Land's guide treats AI SEO as the broader practice with GEO inside it, while other writers use the two as synonyms. GEO as a term comes from a 2023 research paper, published at KDD 2024. ### Do I need a separate strategy for AI search, or is normal SEO enough? For Google, its documentation says specific optimisation is not required for AI Overviews and AI Mode, and the same SEO fundamentals apply. For other AI products, the documented extra step is making sure their search crawlers are allowed in robots.txt. Answer-first writing and testing with real questions are sensible practice, not requirements. ### Can ChatGPT do SEO for me? It can help with tasks such as outlining, drafting and summarising, but it does not check its own facts or take responsibility for what you publish. You still need to review the work and make sure the page genuinely helps the reader. ### Can anyone guarantee my site will be cited in AI answers? No. Google, OpenAI, Anthropic and Perplexity describe eligibility and controls, and none of them promises inclusion. Treat anyone offering a guarantee with caution. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [OpenAI: overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Search Engine Land: What is AI SEO?](https://searchengineland.com/guide/what-is-ai-seo) - [Search Engine Land: SEO, GEO, or ASO? What to call the new era of brand visibility in AI](https://searchengineland.com/seo-geo-aso-new-era-brand-visibility-ai-research-464936) --- # Guide: What is AI visibility? Canonical: https://whatisgeo.app/what-is-ai-visibility/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > AI visibility is how often, and how accurately, AI answer tools mention, cite or describe your business when people ask relevant questions. You can check it and improve it, but nobody can guarantee it. Put simply, when someone asks ChatGPT, Perplexity, Claude or Google's AI features a question your business could answer, AI visibility covers whether you appear in the reply and whether the reply gets you right. The useful way to look at it is in three separate parts, because each one goes wrong for different reasons and needs a different fix. Most pages on this topic are written by vendors and define the term around their own dashboard. This guide splits it into parts you can check yourself. For the wider field it belongs to, see [what GEO is](/guide/#what-is-geo). ## The short definition, and what it covers AI visibility is more than your name turning up in a list. It also covers whether the tool links to your pages as a source, and whether the description is accurate: what you sell, where you work and who you serve. A mention that gets your services or area wrong can send the wrong buyers your way, or put the right ones off, so accuracy counts as much as presence. ## The three parts: being mentioned, being cited, being described correctly You can have any one of these without the others. | Part | What it means | What you check | |---|---|---| | Mentioned | The answer names your business | Is your name in the answer text? | | Cited | The answer links to your pages as a source | Is one of your URLs among the links? | | Described correctly | What it says about you matches reality | Are services, location and anything a buyer would act on right? | You can be cited without being named in the text, and named without being cited. A business that never comes up has a different problem from one that comes up with the wrong service area. Combine them into one number and you lose sight of which problem you have. Our recommendation (this is practice, not something any platform documents): judge every answer on these three parts, and write down which competitors are named alongside you or in your place. ## AI visibility vs SEO rankings: why a score is not a position A search ranking is a position. You are third or tenth for a query, and two people looking at the same results can agree on it. An AI answer is written text put together for that question. It may name five businesses, one or none, so there is no position three to hold. When a tool gives you an AI visibility score, it is summarising something with no fixed scale underneath. SEO still matters. Google's documentation says that to be eligible as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to show in Google Search with a snippet, and that there are no additional technical requirements. It also says specific optimisation is not required for these features and that existing SEO fundamentals still apply. For the longer comparison, see [GEO vs SEO](/guide/#geo-vs-seo). ## Why AI visibility is hard to pin down Google says AI Overviews and AI Mode may use different models and techniques, so the responses and links they show will vary, and that AI Overviews only appear when Google's systems judge them useful, so they often do not trigger. That is a documented reason why one result tells you little. Beyond Google, our working assumption (a recommendation, not a documented rule) is that answers can change between runs, between platforms and when the wording changes. A single screenshot from one tool on one day is an anecdote. Treat AI visibility as a pattern you build up over many questions, tools and dates. Anyone promising a precise, stable figure is claiming more certainty than the thing being measured allows. For how to track it over time, see [measuring progress](/guide/#measurement). ## What affects it: documented requirements, research findings and sensible practice These are three different kinds of evidence. Bad advice often starts when they get mixed up, so here they are separately. ### Documented by the platforms - **Google:** pages must be indexed and eligible for a snippet. Robots.txt rules for Googlebot control crawling for Search, including its AI features. A separate token, Google-Extended, limits training and grounding in some of Google's other systems. - **OpenAI:** OAI-SearchBot is for search. Sites that opt out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links. GPTBot crawls content that may be used for training, and each setting is independent. ChatGPT-User handles certain user-initiated actions, and OpenAI says robots.txt rules may not apply to it. - **Perplexity:** PerplexityBot surfaces and links sites in Perplexity search results, and Perplexity says it is not used to crawl content for AI foundation models. Perplexity-User fetches pages when a user asks a question, and because a user requested the fetch it generally ignores robots.txt. - **Anthropic:** ClaudeBot collects content that may be used for training. Claude-User fetches pages when people ask Claude questions. Claude-SearchBot indexes content to improve search results. Anthropic says its bots honour robots.txt, and that blocking Claude-User or Claude-SearchBot may reduce your visibility in user searches. In practice, blocking a search crawler is a decision about that tool. Check your robots.txt, and any firewall or CDN rules, before you assume something else is wrong. There is more on this under [crawler access](/guide/#crawler-access). Bear in mind that robots.txt (RFC 9309) is a set of rules crawlers are asked to follow. The standard itself says those rules are not a form of access authorisation. ### Research findings Aggarwal et al., "GEO: Generative Engine Optimization" (KDD 2024), built a benchmark of 10,000 queries called GEO-bench. They tested changes to source pages, such as citing sources, adding quotations and adding statistics, against a generative engine they set up: a language model answering from the top five Google results. Visibility was measured with the paper's own metrics, which look at how much of the answer draws on a source and how prominently. The authors report gains of up to 40% on those metrics, say results varied by domain, and also ran tests on Perplexity. This is useful evidence that content choices can matter under test conditions. It does not show that the same changes will get you into a real answer on a real platform today. Treat it as an idea worth testing. ### Sensible practice This is our recommendation and is not sourced: write clear pages that answer the questions your buyers ask, keep your facts consistent across your own site and profiles, and state what you want an AI tool to say about you somewhere it can read. None of this guarantees anything. It is basic housekeeping that also helps human visitors. For how the pieces connect, see [how AI answers happen](/guide/#how-it-works). ## How to check your own AI visibility without a tool You don't need to pay for anything. You need a spreadsheet and an hour. These steps are our recommendation. 1. Write down a fixed set of buyer questions, the ones a real customer would type: what you do, who does it near them, how to choose between providers. 2. Put each question to several tools: ChatGPT, Perplexity, Claude, and Google if an AI Overview or AI Mode answer appears. 3. Record the date, the tool, the exact wording and the full answer. Paste it in rather than summarising from memory. 4. Score each answer: named or not, linked as a source or not, described accurately or not. 5. Note which competitors are named. 6. Repeat on another day with the same questions and compare. People tend to skip step 6, but it is the step that shows whether you are seeing a pattern or a one-off. Run the check again after you change something on your site and you will have your own before and after. For Google specifically, its documentation says traffic from AI Overviews and AI Mode is counted in the Search Console Performance report under the "Web" search type, combined with other search traffic. ## What an AI visibility score does and does not tell you Vendor scores differ from tool to tool. The definitions are proprietary, so you usually cannot see exactly what is counted or how it is weighted. A score can only be compared with other scores from the same tool using the same method: a 60 in one product and a 60 in another measure different things. A score can be a handy way to follow change inside one tool over time. It cannot tell you how you compare across tools, and it cannot promise an outcome. We are not quoting any vendor statistic here, on purpose. Finally, the plainest limit: none of the platform documents we read promises inclusion in an AI answer, and none gives a timeframe. Nobody else can promise either. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you would rather not run the manual check by hand, it tracks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions. - [Hilyt](https://hilyt.it/): If AI tools describe your business wrongly, it gives you an open, AI-readable profile that sets out your facts in one place. ## Questions people ask ### What does AI visibility mean? It means whether AI answer tools such as ChatGPT, Perplexity, Claude and Google's AI features bring up your business when people ask relevant questions, and whether what they say is right. It is easiest to judge in three parts: are you named, are you linked as a source, and are you described accurately. ### How is AI visibility different from SEO rankings? A ranking is a position in a list of results. An AI answer is written text that may name several businesses, one or none, and Google says its AI Overviews and AI Mode can show different responses and links. For Google's features the documented requirement is the same as for Search: be indexed and eligible to show with a snippet. ### How do I check whether ChatGPT or Perplexity mentions my business? Write a fixed list of questions a buyer would ask, put each one to each tool, and record the date, the exact wording and the full answer. Score each answer on named, linked and accurate, then repeat with the same questions on another day, because one answer proves very little. ### Can I guarantee my brand appears in AI answers? No. None of the platform documents we read promises inclusion, and nobody else can promise it either. You can remove blockers such as robots.txt rules, keep your facts clear and consistent, and track what happens. ### What is an AI visibility score? It is a number a vendor's tool calculates to sum up how you appear in AI answers. Each vendor defines and weights its own score, so a score only means something next to other scores from the same tool and method. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full text (HTML)](https://arxiv.org/html/2311.09735v3) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) --- # Guide: What is answer engine optimization (AEO)? Canonical: https://whatisgeo.app/what-is-answer-engine-optimization/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Answer engine optimization (AEO) means making your pages easy for AI assistants to reach, understand and quote. It overlaps almost entirely with GEO and good SEO, and nobody can guarantee a place in an answer. Short answer: AEO is mostly the same work as good SEO, aimed at a new place where the answer appears. The platforms' own documentation asks for very little beyond letting their crawlers in and being indexable. Everything below is labelled as **platform documentation**, a **research finding** or **our recommendation**, so you can see what kind of claim each one is. ## AEO in plain English Someone asks a question. Instead of a page of links, they get a written answer with a few sources attached. AEO is about being one of those sources, or at least being a page the system can reach, make sense of and quote accurately. In practice that means pages that can be crawled and indexed, plain answers near the top, and facts about your business that say the same thing everywhere. If that sounds like good SEO, that is because most of it is. The term is young, and agencies and tool vendors define it slightly differently. Some present it as a new discipline with its own toolkit. Others use it as another name for work you may already be doing. What follows is our reading, and it is a recommendation. ## AEO, GEO and SEO: are they different things? **Our recommendation:** treat them as overlapping labels for the same work. | Term | Usually means | Where the result appears | |---|---|---| | SEO (search engine optimisation) | Being found and ranked in search results | Lists of links | | AEO (answer engine optimization) | Being the source a system quotes when it answers a question directly | AI answers, assistants, answer boxes | | GEO (generative engine optimisation) | Being used and cited when a generative AI system composes an answer | AI answers from ChatGPT, Perplexity, Claude, Gemini, AI Overviews | AEO and GEO describe nearly the same target, and you will see them used interchangeably. Both rest on SEO foundations: a page that cannot be crawled or indexed cannot be quoted. Arguing about the label will not change a single answer, so spend the time on the pages. For the longer version, see [what GEO is](/guide/#what-is-geo) and [how GEO compares with SEO](/guide/#geo-vs-seo). ## What answer engines are and how they produce answers An answer engine replies to a question with a written answer, usually with links to its sources. ChatGPT search, Perplexity, Claude and Google's AI Overviews and AI Mode all work this way. The detail differs by platform, but the broad shape is the same. The system has to reach your page, either through its own index or by fetching it when a user asks. Then it has to read the page and decide whether to use it. **Platform documentation:** Google says AI Overviews and AI Mode may use "query fan-out", issuing multiple related searches across subtopics and data sources to build a response, which lets them show a wider set of supporting links. The mechanics are covered in [how AI answers happen](/guide/#how-it-works). ## What the platforms themselves say you need to do These points are all **platform documentation**, taken from each company's own pages. **Google (AI Overviews and AI Mode).** Google says there are no additional requirements to appear in AI Overviews or AI Mode, and no special optimisations are necessary. A page has to be indexed and eligible to appear in Search with a snippet. You do not need to create new machine-readable files, AI text files or markup, and no special schema.org structured data is required. Inclusion is still not guaranteed. To limit what is shown from your pages, Google points to `nosnippet`, `data-nosnippet`, `max-snippet` and `noindex`. Traffic from these features is counted in your overall Search Console figures, under the "Web" search type. **OpenAI (ChatGPT).** OpenAI runs separate user agents, and each setting is independent. OAI-SearchBot is for search: sites that block it will not be shown in ChatGPT search answers, though they can still appear as navigational links. GPTBot crawls content that may be used to train OpenAI's models, so you can block GPTBot and still allow OAI-SearchBot. ChatGPT-User visits pages when a user asks, and OpenAI says robots.txt rules may not apply to it. OpenAI notes that changes to robots.txt can take about 24 hours to take effect for search. **Perplexity.** PerplexityBot surfaces and links websites in Perplexity's search results, and Perplexity says it is not used to crawl content for AI foundation models. Perplexity-User fetches pages when a user asks a question, and because a user requested the fetch, it generally ignores robots.txt rules. **Anthropic (Claude).** Anthropic lists three bots. ClaudeBot collects content that may be used for training. Claude-User fetches pages when people ask Claude questions. Claude-SearchBot indexes content to improve search results. Anthropic says that blocking Claude-User or Claude-SearchBot may reduce your site's visibility in user search results. The pattern is clear: most platforms separate the bot that finds pages for answers from the bot that collects training data. That gives you a real choice. You can stay visible in answers while opting out of training. robots.txt is the shared mechanism, and it is now a standard (RFC 9309). The RFC itself says its rules "are not a form of access authorization": they ask crawlers to behave, they do not lock anyone out. The practical set-up is in [crawler access](/guide/#crawler-access). **On llms.txt:** llmstxt.org describes it as "a proposal to standardise on using an /llms.txt file" to help AI agents use a website. None of the platform pages above say they require one, and Google says outright that AI text files are not needed for its AI features. Adding one is optional. ## What the research shows, and its limits **Research finding:** the paper that named the field is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024. The authors built GEO-bench, a benchmark of 10,000 queries drawn from several sources, and tested ways of rewriting source content to see which raised its visibility in generated answers. In their tests, adding citations to sources, adding relevant quotations and adding statistics performed best. Keyword stuffing, the classic SEO trick, did not perform well. The abstract reports visibility gains of up to 40%, and the authors found that what works varies by subject area. They also ran the same methods on Perplexity.ai and report gains of up to 37% there. The limits matter as much as the results: - The main experiments used a generative engine the authors built themselves. It was based on GPT-3.5 and fetched only the top five Google results for each query. Today's assistants work differently and change often. - Visibility was measured with the authors' own metrics, some of which were scored by an AI model. - "Up to" figures are best cases under test conditions. They are not forecasts for your site. Read the paper as evidence that content choices can change visibility in AI answers. It shows nothing about what will happen to your pages next month. ## A practical AEO checklist (our recommendations) These are **our recommendations**, based on the documentation and research above. None of them guarantees inclusion. The thinking behind them is in [principles](/guide/#principles). 1. **Let the search crawlers in.** Check your robots.txt, and any firewall or bot protection, for OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot. Decide separately whether to allow the training bots (GPTBot, ClaudeBot). 2. **Be indexable.** Confirm in Google Search Console that your key pages are indexed and can show a snippet. Google says nothing more is required for its AI features. 3. **Answer the question early.** Give a direct, plain answer near the top of the page, then the detail. 4. **Back up claims.** Cite sources and give specific facts. This matches what did well in the GEO study, but only claim figures you can stand behind. 5. **Keep your facts consistent.** Your name, what you do, where you work and your prices should say the same thing on your site and your profiles. 6. **Skip paid shortcuts.** Be wary of anyone selling special markup or files as a route into AI answers. The documentation above does not support it. ## How to tell whether it is working No platform gives you a single report of how often you were quoted. Google counts AI feature traffic within your normal Search Console totals, so you cannot pick it out from there. **Our recommendation:** - Write down the questions your customers actually ask. Put them to ChatGPT, Perplexity, Claude, Gemini and Google on a fixed schedule, and note whether you are named or linked. - Check your server logs for visits from the search bots you allowed, which confirms that access is working. - Watch referral traffic from AI tools in your analytics, alongside your normal search performance. Answers change from one asking to the next, so one appearance tells you little. A pattern over several checks tells you more. More on this in [measuring progress](/guide/#measurement). ## What AEO cannot promise It cannot promise inclusion. Google says plainly that appearing is not guaranteed, and no agency, tool or consultant controls what an answer engine decides to say. It cannot promise a traffic figure either. Be sceptical of any page that quotes one without a source you can open and read. Platforms change how they work and update their documentation to match, so check the sources on this page before relying on any summary of them, ours included. What AEO can do is make your pages easier to reach, read and quote. That is a modest claim, and it is the one the evidence supports. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once your pages are in order, the next question is whether ChatGPT, Claude, Gemini and Perplexity actually name and cite you, and this tracks exactly that. ## Questions people ask ### What does AEO stand for and what is an answer engine? AEO stands for answer engine optimization. An answer engine is any system that replies to a question with a written answer, usually with a few linked sources, instead of only a list of results. ChatGPT search, Perplexity, Claude and Google's AI Overviews and AI Mode all work this way. ### Is AEO the same as GEO? Mostly, yes. Vendors draw the line in slightly different places, and our reading is that the two terms describe the same work from different angles. That is our view of the terminology. There is no official definition. ### Will AEO replace SEO? Anyone who says it will is making a prediction. Google's own documentation says that to appear in AI Overviews or AI Mode, a page must be indexed and eligible to show in Search with a snippet, and that no extra requirements apply. For Google, at least, AEO rests on ordinary search eligibility. ### How do I do answer engine optimization? Let the search crawlers of each platform reach your site, make sure your pages are indexed, answer questions plainly near the top of each page, and keep the facts about your business consistent everywhere. Then check by hand, at regular intervals, whether you appear for the questions your customers ask. ### Do I need special markup or an llms.txt file for AEO? Not for Google: its documentation says you do not need new machine-readable files, AI text files, markup or special schema.org structured data to appear in its AI features. llms.txt is a community proposal, and none of the platform documents cited on this page say they require it. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), abstract](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization, full paper (HTML)](https://arxiv.org/html/2311.09735v3) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [The /llms.txt file proposal](https://llmstxt.org) --- # Guide: What is generative engine optimization (GEO)? Canonical: https://whatisgeo.app/what-is-generative-engine-optimization/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > Generative engine optimisation (GEO) is making your content easy for AI answer systems to find, understand and cite. No one can guarantee inclusion; it is sensible practice plus some early research. In plain terms, GEO asks one question of every important page: could a system that writes answers instead of listing links fetch this page, make sense of it and name it as a source? Most of the work is ordinary good publishing, and so far the formal research behind the term is one benchmark study. ## The short definition, and what it covers Wikipedia describes GEO as "the practice of improving visibility in responses generated by generative artificial intelligence (AI) systems". We use that to check how the word is generally used. It is not evidence that any tactic works. In practice GEO covers three things: - **Access:** whether AI crawlers, and the search indexes behind AI answers, can reach and read your pages. - **Content:** what you say, how plainly you say it, and whether each claim can be checked. - **Consistency:** whether your business is described the same way wherever it appears. It does not cover buying a place in an answer. None of the platform documentation we read for this page describes a paid route into AI answers. Every claim below is one of three kinds, and we say which: a platform's documented behaviour, a named research finding with its limits, or our own recommendation. Where we had no source, we left the point out. The home guide has a shorter version of this definition in [what is GEO](/guide/#what-is-geo). ## Where the term came from The term comes from a paper: Aggarwal et al., "GEO: Generative Engine Optimization", first posted to arXiv in November 2023 and accepted to KDD 2024. The authors present GEO as a way to help content creators make their content more visible in generative engine responses. They also released GEO-bench, a set of 10,000 queries across several domains, for testing methods. The word has since spread well beyond academia, and plenty of agencies use it loosely. When someone quotes a GEO result at you, ask whether it traces back to something you can read. ## How AI answers are produced, and why that matters for GEO Different systems work differently, so we only describe what the platforms themselves document. **Google AI Overviews and AI Mode** are built on Google Search. Google's documentation says that to be shown as a supporting link, "a page must be indexed and eligible to be shown in Google Search with a snippet". It also says "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary". According to the same page, there is no special schema.org structured data to add and no need for new machine readable or AI text files. Both features may use "query fan-out", which means running several related searches across subtopics to build one response. Google also says plainly that "Indexing and serving isn't guaranteed." **OpenAI, Perplexity and Anthropic** each document separate crawlers for separate jobs: | Company | Crawler | What the company says it is for | |---|---|---| | OpenAI | OAI-SearchBot | Showing websites in ChatGPT's search features | | OpenAI | GPTBot | Crawling content that may be used to train its foundation models | | OpenAI | ChatGPT-User | Visiting pages when a user asks; OpenAI says robots.txt rules may not apply | | Perplexity | PerplexityBot | Showing and linking websites in Perplexity's search results; not used to train foundation models | | Perplexity | Perplexity-User | Fetching a page when a user asks; Perplexity says it generally ignores robots.txt | | Anthropic | ClaudeBot | Collecting web content that could contribute to model training | | Anthropic | Claude-SearchBot | Improving the quality of Claude's search results | | Anthropic | Claude-User | Fetching pages when a Claude user asks a question | The practical point is that search and training are controlled separately. OpenAI, for example, says sites that opt out of OAI-SearchBot are left out of ChatGPT search answers. Blocking GPTBot only signals that your content should not be used for training. If you block every AI bot by reflex, you may take yourself out of AI search results as well. Access is the first gate. If a system cannot fetch or index your page, it cannot cite it, and everything else in GEO depends on that. The [crawler access](/guide/#crawler-access) section covers the robots.txt detail, and [how it works](/guide/#how-it-works) gives the wider picture. ## What GEO involves in practice This section is our recommendation. No platform confirms that these steps lead to citations. 1. **Put clear, direct answers near the top of the page.** If a question has a short answer, give it first, then explain. This helps readers whatever the AI systems do; we cannot prove that answer systems favour it. 2. **Name your sources and make claims checkable.** Say where a number came from and link to it. If you cannot point to evidence, soften the claim or drop it. 3. **Keep important text crawlable.** Words that only appear after scripts run may not be seen by every crawler. Plain text in the page is the safer choice. 4. **Describe your business consistently across the web.** Use the same name, description and facts wherever you appear. Most of this is good editorial practice under a new label, which is why it is worth doing even as the AI products change. The [principles](/guide/#principles) section goes through each one in more depth. Which crawlers you let in is your decision, and robots.txt is how you state it. The standard, RFC 9309, says its rules "are not a form of access authorization"; crawlers are "requested to honor" them. Anthropic says its bots honour robots.txt directives. OpenAI and Perplexity say their user-triggered fetchers may not follow it. Decide which bots to allow based on each company's own description of what they do. ## What research has and has not shown The Aggarwal et al. paper is the main named study, so here are the conditions it ran under. - **The engines:** the main tests used the authors' own simulated engine, which took the top five Google results for a query and had gpt-3.5-turbo write the answer. They also tested Perplexity.ai on 200 of the queries. Because Perplexity does not let users choose source URLs, they uploaded the source text as files. - **The methods:** nine ways of rewriting a source page: Authoritative, Statistics Addition, Keyword Stuffing, Cite Sources, Quotation Addition, Easy-to-Understand, Fluency Optimization, Unique Words and Technical Terms. - **The measures:** two visibility metrics the authors designed themselves. One is a word count weighted by where a source is cited in the answer. The other is a "subjective impression" score covering seven aspects. Under those conditions, the paper names Cite Sources, Quotation Addition and Statistics Addition as its top-performing methods. It reports that keyword stuffing offered little to no improvement. The abstract's headline figure is a visibility boost of "up to 40%" on those metrics, and it notes that how well the methods work "varies across domains". The authors say methods may need to change as generative engines evolve, and that the questions people ask change over time. We would add two cautions. This is one study, mostly run on a simulated engine built in 2023 and 2024, and its gains are measured on its own metrics. A result from that setup does not automatically carry over to ChatGPT, Gemini or AI Overviews as they work today. Treat it as early evidence: worth reading, but not a rulebook. ## What GEO cannot do It cannot guarantee you appear in an answer. That is our stance, and the documentation supports it. Google describes eligibility and says serving is not guaranteed, and none of the crawler documentation we read promises inclusion. Being eligible is not the same as being chosen. It cannot make an engine behave predictably, because you do not control the engine. It cannot make up for having nothing worth citing. And nobody outside these companies can tell you for certain how a given model picks its sources, because the platforms do not publish that. If an agency promises a guaranteed spot in ChatGPT or AI Overviews, ask them for the platform documentation that backs it. ## GEO, AEO, LLMO and AI SEO: the names These labels overlap. Wikipedia lists answer engine optimisation (AEO), large language model optimisation (LLMO) and "AI SEO" as other terms for the same concept. In our reading, different writers use them for similar and sometimes identical practices, and no standards body defines any of them. Our working view is that GEO is the only one of these terms with a named research origin, while the others are common in industry writing. For how GEO relates to traditional search work, see [GEO vs SEO](/guide/#geo-vs-seo). We will compare AEO in detail on a separate page. ## Where to start A short, honest list, in this order: 1. Check that your key pages are indexed and can show a snippet in Google Search. That is Google's stated requirement for its AI features. 2. Read each AI company's crawler documentation and decide, bot by bot, what to allow in robots.txt. 3. Rewrite your most important pages so the answer comes first and the sources are named. 4. Make sure the important text is in the page itself, not only rendered by scripts. 5. Make sure your business is described the same way on your site, profiles and listings. The full 30-day version is in [getting started](/guide/#getting-started). To check whether any of it shows up in AI answers, see [measuring progress](/guide/#measurement). ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): Once you know what GEO is, the next question is whether ChatGPT, Claude, Gemini and Perplexity actually name and cite your company, and this measures that. ## Questions people ask ### What is generative engine optimization in simple terms? It means making your content easy for AI answer tools to find, understand and cite. The term comes from a research paper by Aggarwal et al., accepted to KDD 2024. ### How does generative engine optimization work? It comes down to access and clarity. Crawlers and search indexes have to be able to reach your pages, and the content has to be clear and checkable. The platforms document their crawlers and, in Google's case, the eligibility rules, but none of them publish how sources are chosen. ### Is GEO the same as AEO or SEO? Wikipedia lists AEO, LLMO and AI SEO as other terms for the same concept, and none of them has a standard definition. GEO overlaps heavily with SEO. Google says its AI features need nothing beyond a page being indexed and eligible for a snippet. ### Can you guarantee your site appears in ChatGPT or Google AI Overviews? No. Google says indexing and serving are not guaranteed, and none of the crawler documentation we read promises inclusion. Being eligible is not the same as being chosen. ### How do you start with generative engine optimization? Start with access: check indexing and snippet eligibility, then decide bot by bot what robots.txt allows. After that, put answers first, name your sources, and describe your business the same way everywhere. ## Sources - [Aggarwal et al., GEO: Generative Engine Optimization (arXiv abstract, KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization (full paper, v3)](https://arxiv.org/html/2311.09735v3) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://developers.openai.com/api/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic: Does Anthropic crawl the web and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-the-web-and-how-can-site-owners-block-the-crawler) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Wikipedia: Generative engine optimization](https://en.wikipedia.org/wiki/Generative_engine_optimization) --- # Guide: What is llms.txt, and does it actually work? Canonical: https://whatisgeo.app/llms-txt/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > llms.txt is a proposed Markdown file that points AI agents to your key pages. It is not a web standard and no major AI search platform documents using it to pick citations, so it is cheap to add but unproven. An llms.txt file is a plain Markdown index of your site, written for AI agents that fetch pages while they help someone. Whether it gets you cited by ChatGPT, Perplexity or Google's AI features is unproven: none of those platforms documents reading it for that purpose, and the largest analysis we found saw no link between having one and being cited. Below, each claim is marked as one of three kinds: what the proposal says, what platforms have documented, and what is only our recommendation. ## What llms.txt is, and who proposed it The proposal lives at llmstxt.org. Jeremy Howard wrote it, first published it on 3 September 2024 and revised it as version 2 in August 2026. It starts from the point that web pages are built for people: navigation, ads and scripts get in the way when an agent wants clean text, and most sites are too big to fit in a model's context window. So the proposal suggests one small Markdown file that gives agents a short overview and links to the detail. Three points from the proposal itself are worth knowing: - **Where it goes.** At `/llms.txt`, or at any path such as `/docs/llms.txt`. A file covers the pages under its path, and where more than one applies, agents should use the most specific. - **Markdown copies of pages.** It also proposes that useful pages offer a clean Markdown version at the same URL, either with `.md` added (`page.html.md`) or with the extension swapped (`page.md`). Version 2 adds standard link relations so agents can find them: `rel="alternate" type="text/markdown"` for a page's Markdown copy and `rel="describedby"` for the llms.txt file that covers it. - **How it is meant to be read.** Agents are expected to view or search the file, then follow the relevant links. Version 1 described a tool for expanding the file into one big context; version 2 drops it. The proposal text we read does not define a separate llms-full.txt file. Its status matters. The proposal calls itself an informal overview, open for community input and hosted on GitHub. It has not been through a standards body. Compare robots.txt, which is an IETF Standards Track document, RFC 9309. ## What the proposal says the file should contain The format is small and ordered. In sequence, a file contains: 1. An optional byte order mark. 2. An H1 with the name of the project or site. This is the only required part. 3. A blockquote with a short summary holding the key information needed to understand the rest. 4. Zero or more Markdown sections of any type except headings (paragraphs, lists) with more detail. 5. Zero or more sections under H2 headings, each a "file list": a Markdown list where every item has a `[name](url)` link, optionally followed by `:` and a note. By convention, a section called "Optional" holds secondary links an agent can skip when it needs a shorter context. Since version 2 it carries no special mechanical meaning; it is simply a useful label. ## An annotated llms.txt example This is our own illustrative sample for a made up business. It is an example and does not describe any real site. ```markdown # Harbour Bakery > Independent bakery in one location selling bread, pastries and > celebration cakes. Custom cake orders need two days' notice. Prices and opening hours change seasonally; the linked pages are the current source. ## Key pages - [Menu](https://example.com/menu.md): current bread and pastry list with prices - [Custom orders](https://example.com/orders.md): how to order a cake and the notice needed - [Opening hours](https://example.com/hours.md): weekly hours and holiday closures ## Optional - [Our story](https://example.com/about.md): background on the bakery ``` Reading it line by line: - **`# Harbour Bakery`** is the H1, the one required element. - **The blockquote** is the summary an agent reads first, so it carries the facts that matter most. - **The plain paragraph** is the optional detail section. Here it tells the agent which pages to trust for things that change. - **`## Key pages`** is a file list. Each link has a note after the colon so an agent knows what the page holds before fetching it. The links point to `.md` copies, as the proposal suggests; if your site has no Markdown versions, ordinary page URLs still fit the format. - **`## Optional`** holds material an agent can skip. Notice what the file does not do. It grants no permissions, blocks nothing and carries no ranking signal. It only describes. ## llms.txt vs robots.txt vs sitemap.xml | File | What it does | Status | Does anything have to follow it? | |---|---|---|---| | robots.txt | Tells crawlers which parts of a site they may access | IETF standard, RFC 9309 | Crawlers that follow the standard honour it, though the RFC says its rules "are not a form of access authorization" | | sitemap.xml | Gives search engines details about your pages | Sitemaps protocol (sitemaps.org) | No; the protocol says search engines "may use" the information | | llms.txt | A curated overview for agents, used on demand while answering | Informal proposal | No; nothing obliges any tool to read it | The proposal itself says llms.txt is meant to sit alongside the other two: a sitemap lists everything indexable, while llms.txt is a short, curated path that can also include external links. If your real question is whether AI crawlers can reach your pages at all, that is a robots.txt and crawler question, covered in [crawler access](/guide/#crawler-access). ## Is llms.txt actually used? What is documented and what is not "Used" can mean three different things, and the evidence differs for each. **Sites publishing the file.** The proposal's author says thousands of sites now publish one. The proposal lists documentation platforms and CMS tools that generate the file automatically, including Mintlify, GitBook, Yoast SEO and Wix. That adoption figure is the author's own claim. AI companies publish llms.txt files for their own developer documentation; OpenAI's and Perplexity's crawler documentation pages both point readers to an llms.txt index. The proposal says the heaviest use is software documentation, where coding agents follow the files to find API references. **Tools checking for it.** Chrome's Lighthouse has an llms.txt audit among its agentic browsing checks. It describes the file as "an emerging convention" that gives LLMs and AI agents a machine readable summary of a site. The audit fails only if the server errors when Lighthouse requests the file. A 404 makes the audit not applicable, because the file is "optional at the moment". The page does not say that any AI system reads the file. **AI answer systems reading your file to decide what to cite.** This is where the documentation runs out. We read each company's crawler documentation: - **OpenAI** describes OAI-SearchBot (ChatGPT search), GPTBot (training) and ChatGPT-User (user actions), all managed through robots.txt. It does not mention llms.txt as something its crawlers read. - **Anthropic** describes ClaudeBot, Claude-User and Claude-SearchBot, also controlled through robots.txt. No mention of llms.txt. - **Perplexity** describes PerplexityBot and Perplexity-User, with robots.txt and published IP ranges. No mention of llms.txt for crawling. Absence from documentation is not proof that nobody reads the file. An agent fetching pages for a user could open it. What we can say is that none of these companies documents using it, so any claim that they do needs its own source. ## Does llms.txt work? What the evidence does and does not show The best evidence we found is a vendor study. SE Ranking, an SEO software company, analysed nearly 300,000 domains and found that 10.13% had an llms.txt file. It then compared that with how often each domain appeared as a source in sampled LLM responses. Using Spearman correlation, an XGBoost model and SHAP analysis, it reported no correlation between AI citations and having the file. It also found its model became more accurate when the llms.txt variable was removed. Read that with its limits. It is SE Ranking's own research. It is observational: sites chose whether to add a file, and nobody tested before and after. The article we read does not name which AI systems were sampled, and it calls its findings context dependent. It does not prove the file has no effect. What it does mean is that we could find no published evidence of a benefit. The best known research on generative engine optimisation, Aggarwal et al.'s GEO paper (accepted to KDD 2024), does not settle the question either way. It tested ways of changing page content on its own benchmark, GEO-bench, and reported visibility gains of up to 40% in that setup. It did not test llms.txt. We found no controlled experiment with published results showing that adding llms.txt changes citations. ## llms.txt and Google Google's guidance on AI features in Search (AI Overviews and AI Mode) is direct: "You don't need to create new machine readable files, AI text files, or markup to appear in these features." It also says there is no special schema.org structured data to add. To be eligible, a page must be indexed and eligible to be shown in Google Search with a snippet, and "there are no additional technical requirements". The page does not name llms.txt, but its wording covers it, since it says no AI text file is needed. So for Google, the documented route is ordinary indexing, which is the same ground covered in [GEO vs SEO](/guide/#geo-vs-seo). ## Should you add one? A sensible-practice checklist This section is our recommendation; no platform requires any of it. For most small sites an llms.txt file is a short job and costs nothing to host, so adding one is reasonable as long as you expect nothing measurable from it. 1. **Serve it at `/llms.txt`** as plain text and check it returns a 200 status. A 404 is harmless; a server error is what fails the Lighthouse audit. 2. **Start with the H1**, then a blockquote summary with the facts a stranger most needs (what you do, where, for whom). 3. **List only the pages you would want read first**, each with a one line note after a colon. 4. **Use full, canonical URLs** that resolve without redirects or logins. Click every one. 5. **Put secondary links under `## Optional`.** 6. **Keep it in step with the site.** Wrong opening hours or a dead link in an llms.txt file is worse than having no file. 7. **Check that robots.txt is not blocking the pages you list.** The file cannot help an agent reach pages it is not allowed to fetch. 8. **Test it the way the proposal suggests:** give an AI assistant only your llms.txt and ask it questions about your business. If the answers are wrong, fix the summary. 9. **Do not pay for a service that promises citations from it.** Nobody can promise that. Writing the file has one benefit you can count on: it forces you to decide which pages matter and to say in two lines what you do. That clarity is worth having on your actual pages too. ## What to do first if AI visibility is the goal Do the documented things before the speculative one. 1. **Make sure the right crawlers can reach you.** Every platform above documents robots.txt controls, so start with [crawler access](/guide/#crawler-access). 2. **Understand how assistants find and use pages**, explained in [how AI answers happen](/guide/#how-it-works). 3. **Get the content right.** Clear, specific, well structured pages are what the GEO research worked on; see [principles](/guide/#principles). 4. **Measure.** Check how AI assistants describe you before and after any change, including adding llms.txt, so you work from your own results instead of someone else's claim. [Measuring progress](/guide/#measurement) covers how. llms.txt can sit alongside all of that. It should not come first. ## From the same studio These are made by the studio that publishes this guide. - [Hilyt](https://hilyt.it/): If you want AI tools to find an accurate, readable description of your business beyond a single llms.txt file, Hilyt publishes open AI-readable profiles. - [visibilitytoai](https://visibilitytoai.com/): To find out whether adding llms.txt or anything else changes how ChatGPT, Claude, Gemini and Perplexity name and cite you, VisibilityToAI measures it. ## Questions people ask ### What is an llms.txt file? It is a Markdown file at /llms.txt, or at a subpath, that gives AI agents a short overview of a site and links to the detail. Under the llmstxt.org proposal, an H1 with the site name is the only required part. ### Is llms.txt still used, and by whom? Many sites publish one, including AI companies for their own developer docs, and documentation platforms generate it automatically. Chrome's Lighthouse checks for it. None of OpenAI, Anthropic or Perplexity documents its crawlers reading it to decide citations. ### Does llms.txt really work for getting cited by ChatGPT or Perplexity? We found no controlled evidence that it does. An observational study by the SEO vendor SE Ranking, covering nearly 300,000 domains, found no correlation between having the file and being cited by AI systems. ### Does Google use llms.txt? Google's guidance for AI Overviews and AI Mode says you don't need to create new machine readable files, AI text files or markup to appear in them. A page needs to be indexed and eligible to show with a snippet, and Google says there are no additional technical requirements. ### Is llms.txt needed, and how is it different from robots.txt? It is not needed. robots.txt is an IETF standard (RFC 9309) that tells crawlers what they may access. llms.txt is an informal proposal that only describes your site, and no tool is obliged to read it. ## Sources - [llmstxt.org: The /llms.txt file, v2 (Jeremy Howard)](https://llmstxt.org/) - [llmstxt.org: Changes in v2](https://llmstxt.org/changes.html) - [Chrome Lighthouse: llms.txt audit](https://developer.chrome.com/docs/lighthouse/agentic-browsing/llms-txt) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [Sitemaps.org: Sitemaps XML format](https://www.sitemaps.org/protocol.html) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Anthropic: Does Anthropic crawl data from the web?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [SE Ranking: Does llms.txt impact AI visibility and citations? (vendor research)](https://seranking.com/blog/llms-txt/) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) --- # Guide: Which AI optimisation software improves visibility the most? Canonical: https://whatisgeo.app/which-ai-optimization-software-improves-visibility/ Publisher: Adapt Progress Evolve Limited Updated: 2026-11-10 > No independent evidence ranks any AI optimisation software by how much visibility it adds. Most tools measure visibility; the improvement comes from changes to your site and content, which you can test for yourself. Software can help you see your AI visibility and can speed up the work of changing it. What it cannot show you is proof that the software itself caused an improvement. This page explains where tools help, what the evidence for any change actually rests on, and how to test any product yourself, ours included. None of it requires buying anything. ## Is there a tool that improves AI visibility the most? Nobody can honestly name one. We know of no independent study that compares products by the visibility they added to the same sites, from the same starting point, over the same period. Without a study like that, "the most" has nothing to be measured against. What can be said is narrower. Some tools are good at showing you what AI assistants say about you. Some help you produce or restructure content faster. Neither of those, on its own, tells you that your visibility rose because of the software. ## Measuring visibility and changing it are different jobs A tracker watches. You give it prompts, it records which brands and sources appear in the answers, and it shows you a trend. That is useful, and [how to measure AI visibility](/how-to-measure-ai-visibility/) covers it in detail. But a thermometer does not lower your temperature. An optimisation tool, in the loose sense the market uses, suggests or makes edits: rewrites, structural changes, schema suggestions, content briefs. The edits are what might make a difference. The software is how they get delivered. Plenty of products do both jobs under one login, which is fine. The trouble starts when the measuring half gets the credit for the changing half. A dashboard showing mentions going up tells you that mentions went up. It does not tell you why. For an overview of the categories of product available, see [generative engine optimisation tools](/generative-engine-optimization-tools/). That page does not rank products by results either. ## What actually moves visibility, and what each claim rests on The evidence comes in two kinds: what the platforms document, and what one research paper found. Treat them differently. ### What the platforms document **Google.** Google's guidance on AI Overviews and AI Mode says you can apply the same foundational SEO best practices as for Search overall. To be eligible as a supporting link, a page must be indexed and eligible to be shown in Search with a snippet, and Google states there are "no additional technical requirements". It also says specific optimisation is not required for these features. Any tool claiming a special method for Google's AI features is claiming something Google's own documentation does not support. **Crawler access.** Whether an AI system can reach your pages is something you control, and the platforms document it: | Platform | Crawler | What the documentation says | | --- | --- | --- | | OpenAI | OAI-SearchBot | Used to surface sites in ChatGPT's search features. Sites that opt out will not be shown in ChatGPT search answers, though they can still appear as navigational links. | | OpenAI | GPTBot | A separate setting, for indicating whether content may be used to train OpenAI's foundation models. | | Perplexity | PerplexityBot | Used to surface and link sites in Perplexity search results; Perplexity recommends allowing it. | | Perplexity | Perplexity-User | Fetches pages when a user asks a question. Perplexity says it generally ignores robots.txt because a user requested the fetch. | | Anthropic | Claude-SearchBot, Claude-User | Anthropic says blocking either may reduce your site's visibility in user search results or user-directed web search. | | Anthropic | ClaudeBot | Collects content that may be used for training; blocking it signals that future material should be excluded. | Checking these in your robots.txt costs nothing, and the platforms state the effect themselves. Google's same guidance also lists making sure crawling is allowed in robots.txt "and by any CDN or hosting infrastructure", so check your firewall too. For more detail, see [AI crawlers and robots.txt](/ai-crawlers-robots-txt/). ### What the research found Aggarwal et al.'s GEO paper (accepted to KDD 2024) is the most cited study of content changes for AI answers. Its conditions matter: - It used GEO-bench, a benchmark the authors built of 10,000 queries, about 80% of them informational. - The engine they tested was their own two-step setup: fetch the top five Google results for a query, then have gpt-3.5-turbo write an answer from them. They sampled five responses per query to reduce statistical variation. - Visibility was scored with the authors' own metrics: a position-adjusted word count of the cited text, and a "subjective impression" score rated by GPT-3.5. Under those conditions, the authors report that adding citations, quotations and statistics improved the position-adjusted metric by 30 to 40% relative to unchanged content, while keyword stuffing gave little or no improvement. They also report improvements of up to 37% in a test on Perplexity.ai, and they found that results varied by domain. Read it for what it is: a controlled experiment on one benchmark, using a 2023 model and metrics defined by the authors. It does not show that the same edits will lift your brand in live ChatGPT answers today, and it tested no commercial software. It supports making certain kinds of content changes. It says nothing about which tool to buy. ## Why 'best tool' rankings cannot answer this Most of the top results for this question are lists, often published by a vendor or by an affiliate paid per sign-up. Many say "we tested" without saying how. A ranking only means something if it states a method. Which prompts? How many runs? Which sites, starting from what baseline, changed in what way, over what period? Without those answers, a ranking is an opinion with a number attached. Variation is a problem underneath all of this. Google says AI Mode and AI Overviews may use different models and techniques, so the responses and links they show will vary, and that AI Overviews often do not appear at all. The GEO researchers sampled every answer five times for the same reason. You can see it yourself by asking an assistant the same question several times in fresh chats and comparing the answers. If the answer changes when you repeat the question, one snapshot before and one after proves very little. Measurement can still tell you a lot. It just has to be repeated and controlled before it means anything. ## How to test whether a tool helped you You can run a fair test on any product. It takes discipline more than money. 1. **Fix your prompt set before you start.** Write down the questions your customers would plausibly ask. Do not change them mid-test or add the flattering ones afterwards. 2. **Record a baseline, repeatedly.** Run each prompt several times over several days, in each assistant you care about. Note whether you are named, how you are described and which sources are cited. The repeats show you your normal spread. 3. **Change one thing at a time.** If you rewrite fifty pages, fix robots.txt and publish a new guide all in one week, you will not know which change did what. Make one change, log the date and leave it alone. 4. **Keep a control.** Leave a comparable set of pages and prompts untouched. If both groups move, the tool is not what moved them. 5. **Measure again the same way.** Same prompts, same number of repeats, same assistants. Compare the shift with your baseline spread. 6. **Check the cause.** Did the cited page actually change? Did a crawler start reaching it? If you cannot point to a clear cause, be wary of giving the software the credit. The test works however you do the recording, whether in a spreadsheet or in a tool. For the changes themselves, start with [how to improve AI visibility](/how-to-improve-ai-visibility/). ## Questions to ask a vendor before you pay You do not need to be adversarial, just to get straight answers. 1. Does the product measure, change content, or both? Which feature does which? 2. What is the method behind any claim of improvement: which sites, what baseline, what time window? 3. How does it handle answers varying between runs? How many times is each prompt run? 4. Is the evidence your own or independent? If it is your own, where is it published? 5. Can I export the raw prompts, answers and citations so I can check them without your dashboard? 6. What happens to my data and history if I cancel? Treat any statistic a vendor quotes as the vendor's own claim, made on its own terms. How tools collect their data also differs a great deal; [AI search optimisation tool data accuracy](/ai-search-optimization-tool-data-accuracy/) covers what to check. ## What to do first if you have a small budget Put in your own time before you spend money. Check that the crawlers you want can reach your site, using the platform documentation above. Get the ordinary SEO basics right, since Google says those are what its AI features rely on. Write your prompt set and run a baseline by hand; a spreadsheet will do. Then change one thing and watch what happens. Software can come later, once you know which question you need it to answer. Saving time on recording and reporting is a fair reason to buy a tool. "It will raise our visibility" is a claim you cannot check before you pay, so do not buy on that promise alone. The home guide's section on [measuring progress](/guide/#measurement) has more on setting up a routine you can repeat. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want the recording step of the test done for you, it runs buyer questions across ChatGPT, Claude, Gemini and Perplexity and records whether your company is named and cited; put the same vendor questions to it as to any other tool. ## Questions people ask ### Which AI optimisation software improves visibility the most? Nobody can honestly name one. We know of no independent study that compares products by the visibility they added to the same sites under the same conditions. Test any tool on your own site with a fixed prompt set, a repeated baseline and an untouched control. ### Can software guarantee my brand appears in ChatGPT or AI Overviews? No. Google says a page has to be indexed and eligible for a snippet to appear as a supporting link, and that AI Overviews often do not show at all. Being eligible is no promise of inclusion, and no vendor controls what an assistant chooses to say. ### What is the difference between an AI visibility tracker and an optimisation tool? A tracker asks AI assistants questions and records whether you are named or cited. An optimisation tool suggests or makes changes to your content. Many products do both, so ask which feature does which. ### How do I know if an AI visibility tool is working? Fix your prompts, record a repeated baseline, change one thing at a time, keep an untouched control and then measure again the same way. Credit the tool only if the change is bigger than your normal run-to-run spread and you can point to the cause. ### What are the best AI visibility analytics for search optimisation? Look for tools that show you the raw prompts and answers, run each prompt more than once, list the cited sources and let you export the data. We do not name a winner, because there is no independent method to rank them. ## Sources - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), arXiv:2311.09735](https://arxiv.org/abs/2311.09735) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Aggarwal et al., GEO: Generative Engine Optimization, full text (arXiv HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: Which AI search optimisation tool has the most accurate data? How to test any of them Canonical: https://whatisgeo.app/ai-search-optimization-tool-data-accuracy/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-21 > No independent test names one AI search optimisation tool as most accurate, so be wary of anyone who does. Judge any tool by how it samples answers, which surface it measures, and whether you can reproduce its numbers. Any ranking that names one tool the most accurate needs a public method behind it, and the pages that rank for this question don't show one. What you can do is test any tool yourself in an afternoon, using your own prompts, your own manual checks and your own analytics. ## Why nobody can honestly name the most accurate tool Search this question and you'll mostly find vendor listicles. Each one puts itself, or a partner, at the top. None publishes a test set, a ground truth or a repeat run. A numbered list with no method behind it is marketing. The vendors may be telling the truth about their own product. The problem is structural. There is no shared benchmark for AI visibility data, and no independent body running the same prompts through every tool and checking the results against a known answer. Without that, "most accurate" has nothing to be measured against. So change the question from which tool wins to how you would check any of them. If you want to see what is on offer first, our [generative engine optimisation tools](/generative-engine-optimization-tools/) guide covers the field. This page covers only how to test the data. ## What 'data accuracy' means for AI visibility tools In a classic rank tracker, accuracy is simple: a page was at position four or it wasn't. AI answers don't work like that. An AI visibility tool reports on sampled answers that vary. It asks a model a question, reads the reply and records whether your brand was named or cited. Change the wording, the model, the moment or the surface and the answer can change too. So "accurate" breaks into three separate questions: 1. Did the tool record the answer the model really gave? 2. Does that answer reflect what your buyers would see? 3. Would the tool get a similar result if it ran the same prompt again? A tool can do well on the first and badly on the second. That is why a single headline score tells you very little. ## Where the numbers can go wrong **Repeat runs.** The model providers say this themselves. Anthropic's documentation states that even with temperature set to 0, results "will not be fully deterministic" and identical inputs may produce different outputs across API calls. OpenAI's guidance on its seed parameter says its APIs are non-deterministic by default and that "determinism is not guaranteed". In practice, one run is one sample. Even the original GEO research treated answers this way. Aggarwal et al. (KDD 2024) generated answers with gpt-3.5-turbo from the top five Google results for each query, and sampled five responses at temperature 0.7 "to reduce statistical deviations". Their reported gains (up to 40% on their own benchmark, and up to 37% in a test on Perplexity) are measured against that setup and their own visibility metrics, and they found that results varied by domain. Read the paper for its method. Don't treat its numbers as a forecast for your site or for any tool. **The surface.** A tool might query a model through an API, load a logged-out web interface or read Google's AI features. These are different things. OpenAI's help centre explains that ChatGPT search can rewrite your query, use your approximate location and draw on saved memories. A clean API call won't reproduce all of that, and a tool that mixes surfaces without telling you gives you a number you can't interpret. **The prompt set.** If the vendor picks the prompts, you are measuring the vendor's view of your market. Your buyers may ask quite different questions. **The answer itself.** OpenAI warns that search results and citations "can be incomplete, outdated, or incorrect". A tool can record an answer faithfully and that answer can still describe you wrongly. That counts as a finding about the assistant, and the tool should show it to you. **Google's reporting.** Google documents that sites appearing in AI Overviews and AI Mode are counted in Search Console's Performance report as part of overall search traffic, within the Web search type. Its AI features page does not describe a separate AI-only view in that report. If a tool claims to split out AI Overview traffic using Search Console data alone, ask how, because Google's documentation doesn't describe a way to do it. ## A test you can run on any tool before you pay This is our recommendation. No platform sets these rules. 1. **Write your own prompt set.** Use the questions your buyers actually ask, mostly without your brand name in them. Record the exact wording and the date. 2. **Run them by hand.** Put each prompt into the assistants you care about, more than once. Note whether you are named, whether you are cited and whether the description is correct. 3. **Load the same prompts into the tool.** If it won't accept your prompts, you have learned something already. 4. **Compare answer by answer.** Where the tool says you appeared and your own runs say you didn't (or the other way round), open the raw answer. If the tool shows only a score, you can't audit it. 5. **Repeat the tool's run.** Run its check several times over a few days. If its own results swing a lot between runs and it reports only one figure, that figure is hiding the spread. 6. **Cross-check against analytics.** Look for visits from the assistants in your referral data. For ChatGPT, links have carried a `utm_source=chatgpt.com` tag. OpenAI's help centre doesn't document it, though, and Search Engine Roundtable reported the tag going missing in October 2026 before it was found to be a bug and restored. Treat tagged visits as a minimum count. Analytics only catches people who clicked, and many AI answers influence a buyer without sending a visit. Still, if a tool's numbers never line up with anything you can see, question them. Our guide on [how to measure AI visibility](/how-to-measure-ai-visibility/) goes deeper on method, and the [AI visibility checker](/ai-visibility-checker/) guide covers quick manual checks. ## What each kind of tool can and cannot be accurate about | Method | What it can be accurate about | What it can't promise | |---|---|---| | Queries a model via API | Consistent logging of what that model returned for that prompt | That the answer matches the consumer app your buyers use | | Loads a logged-out web interface | Closer to what a person sees | What logged-in users with memory or location see, or any moment other than the run | | Reads Google surfaces or Search Console | What Google exposes | An AI-only split that Google's own reporting doesn't provide | None of these is wrong in itself. Each one is accurate about something narrower than its marketing may suggest, and your job is to find out what. If Google is your main concern, see [how to track AI Overviews](/how-to-track-ai-overviews/). ## Red flags in tool marketing 1. A single "visibility score" with no definition of how it is built. 2. No mention of how many times each prompt is run. 3. "100% accurate" claims, or any accuracy figure without a method. 4. No way to see the raw answers behind a number. 5. A ranking page where the author's own product comes first. 6. Statistics presented as industry fact when they come from the vendor's own data. On that last point, any vendor statistic belongs to the vendor and should be attributed to them. Our [GEO statistics](/generative-engine-optimization-statistics/) page shows how to separate research findings from vendor claims. Before you sign up, ask each vendor these questions (our recommendation): - Which surfaces do you measure: an API model, a logged-out web interface or Google surfaces? - How often do you run checks? - How many runs per prompt, and do you show the spread? - Can I use my own prompts? - Can I see the raw answers? A vague reply tells you something too. ## What to do with imperfect data All of this data is imperfect, because it is built on sampled answers that vary, and no tool gets around that. Use the numbers to see which way things are moving and don't read them as exact counts. Run the same prompt set several times and report the spread as part of the result. Keep the dates and exact prompts on file so next month compares like with like. Use your manual checks as the reference point, and let a tool earn your trust by matching them. Look at trends across repeated runs over several weeks, not single readings. A brand named more often across many runs has shown you something. One lucky run hasn't. For how this fits into a wider measurement routine, see the [measurement section](/guide/#measurement) of our main guide. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want a tool to put through this test, it runs buyer questions across ChatGPT, Claude, Gemini and Perplexity and records whether your company is named and cited, and you should check it the same way as any other. ## Questions people ask ### Which AI search optimisation tool is the most accurate? No independent test settles it, so no honest source can name one. A better question is whose data you can reproduce. Run the test on this page on two or three tools and trust the one that matches your own checks and shows you its raw answers. ### Why do AI visibility tools give different numbers for the same brand? They use different prompts, measure different surfaces and run each prompt a different number of times. Model providers also say identical inputs can produce different outputs, so even two runs of the same tool can differ. Different numbers usually mean different methods, not a broken tool. ### Are tools that query an API as accurate as checking ChatGPT myself? Not necessarily. An API is repeatable and easy to log, but it may not match what a person sees in the ChatGPT app, where search, location and memory can shape the answer. Treat your own checks as the reference and ask the vendor exactly which surface they measure. ### How do I check whether a tool's citation data is right? Open the raw answer behind each citation and compare it with your own run of the same prompt, noting the date. If the tool can't show you the answer, you can't verify the claim. ### Can I trust vendor statistics about AI visibility? Treat them as the vendor's own claims, because that is what they are. Look for a stated method, a sample size and a date, and attribute the figure to the vendor wherever you repeat it. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Anthropic (Claude docs): Glossary, temperature](https://docs.claude.com/en/docs/about-claude/glossary) - [OpenAI Cookbook: How to make your completions outputs consistent with the seed parameter](https://cookbook.openai.com/examples/reproducible_outputs_with_the_seed_parameter) - [OpenAI Help Center: Searching the web with ChatGPT](https://help.openai.com/en/articles/9237897-chatgpt-search) - [Search Engine Roundtable: OpenAI drops UTM parameters from ChatGPT results (October 2026, updated)](https://www.seroundtable.com/utm-parameters-missing-chatgpt-42299.html) - [Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024 (arXiv)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: full paper (HTML, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: Which AI search optimisation tool is most intuitive? How to test any of them yourself Canonical: https://whatisgeo.app/which-ai-search-optimization-tool-is-most-intuitive/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-22 > We found no independent test that ranks AI search optimisation tools for ease of use, so any 'most intuitive' list is opinion. You can judge it yourself with a fixed task list and a free trial or demo. Ease of use depends on who is using the tool and for what, so the only verdict worth having is one you reach on your own tasks. Below is a test you can run on any tracker in about an hour, with a task list, a scoring sheet and notes on how different kinds of tool tend to behave. ## Is there an independent ranking of the most intuitive tool? We did not find one. The pages that come top for this question are mostly vendor pages and affiliate roundups. They call tools intuitive without describing a task, a tester or a score, so "most intuitive" in those sources is the publisher's opinion. An opinion from someone who has used a tool for months still has some value. You just can't audit it, so you can't tell whether it fits your team. ## What "intuitive" means for this kind of tool It is subjective. Software that feels obvious to someone who works in SEO tools all day can baffle a shop owner, and a simple tool can frustrate a specialist who wants the raw data. So split the word into things you can observe: 1. Can you get set up without reading the documentation? 2. Can you enter the questions you want checked without guessing at the format? 3. Does the first report tell you something you can act on? 4. Can you get your data out in a form you can use? In practice, a tool is intuitive for you if you reach a useful answer quickly and without help. The test below checks exactly that. ## A one-hour usability test you can run on any tool This is our recommendation. No published standard exists for testing these tools, and the hour is a suggestion: shorter is fine for a quick screen, longer if two tools come out close. 1. **Get access.** Start a free trial or book a demo. If the only option is a demo, ask to drive it yourself. 2. **Fix your inputs.** Use the same brand, the same competitors and the same set of questions your customers would ask, in every tool. If the inputs change, you are comparing the inputs rather than the tools. 3. **Build a scoring sheet.** List the tasks below down the side and give each tool a column. Score each task from one to five, and add a notes column for every point where you got stuck or had to look for help. The notes usually tell you more than the totals. 4. **Test as the real user.** Do it yourself, or watch the person who will use the tool every week do it. A vendor or specialist demoing the product shows you how it works for an expert, and that may not be how it works for you. ## Tasks to give every tool Keep the list the same for every tool. | Task | What to do | What to note | |---|---|---| | Setup | Create the account, add your brand and competitors, choose location and language if asked | How many screens it took, anything unclear, anything the tool chose for you | | Prompt entry | Add the questions you want tracked | Whether it suggests prompts, accepts a pasted batch or makes you type one at a time | | First report | Find what the tool says about your brand without help | Whether you can read it cold, and whether mentions and citations are shown separately | | Repeat check | Look for answers to the same prompt from different runs or dates | Whether you can see the variation or only a single score | | Export | Get the data out | Whether you get rows (date, engine, prompt, answer) or only charts and screenshots | | Explain it | Tell a colleague what the tool found in two sentences | If you can't, the report isn't doing its job | The repeat check matters because AI answers are not fixed. Google, for example, says AI Mode and AI Overviews "may use different models and techniques, so the set of responses and links they show will vary." A tool that hides that variation behind one number is easier to read but tells you less. Our guide to [how to measure AI visibility](/how-to-measure-ai-visibility/) explains why repeat runs matter. ## What differs by kind of tool: add-ons to an SEO suite, dedicated trackers, and doing it by hand The kind of tool shapes the experience more than any single feature. These are descriptions, not a ranking. **Add-ons to an SEO suite.** Some established SEO platforms now include AI visibility features. Semrush's help centre says its AI Visibility Toolkit is "designed for SEOs and marketers" and built "with SMBs, agencies, and mid-market companies in mind". It also says some reports are tied to the location and language you pick during setup, that competitors for its visibility score are identified automatically, and that data can be exported as CSV. Ahrefs describes Brand Radar as a way to track how your brand appears in AI answers. You can add your own prompts or have AI generate them, and choose to track them daily, weekly or monthly, and there is also a prebuilt index that needs no setup. The upside of a suite is familiarity if you already use it. The catch is that you inherit its complexity, so if the main product already feels heavy, test the add-on with that in mind. **Dedicated trackers.** These do one job: running prompts against AI engines and recording whether you are named or cited. Because they do less, setup is often shorter, but a narrow tool may assume you already know what to do with the results. Our overviews of [generative engine optimisation tools](/generative-engine-optimization-tools/) and [AI SEO tools](/ai-seo-tools/) set out what each kind does. **Doing it by hand.** Ask the questions yourself, record what comes back and repeat. There is no software to learn, and it costs nothing but time. It gets slow once you track many prompts, engines or competitors. Our [AI visibility checker](/ai-visibility-checker/) guide shows how to start this way. ## Who it suits: small business owner, marketer, in-house SEO The same tool can get different verdicts from different people, so score it as the person who will use it. - **Small business owner:** you have little time and may not know SEO terms. The cold read matters most: can you open the first report and understand it without help? If you need a glossary, the tool probably isn't for you. - **Marketer:** reporting matters most. Can you export something clean for a deck, and compare periods and competitors without rebuilding the view each time? - **In-house SEO:** control matters most. You will want to manage prompts, see repeat runs and get the raw answers. A tool that hides the detail behind one score will frustrate you, however tidy it looks. ## Easy to use is not the same as accurate or useful A smooth interface tells you nothing about whether the numbers are right. A tool can be pleasant to use and give you a thin picture, and an awkward one can collect its data well. Run the usability test first, then check accuracy separately: ask which engines it queries, how many runs sit behind each number and where the prompts come from. Our page on [AI search optimisation tool data accuracy](/ai-search-optimization-tool-data-accuracy/) covers those questions, and [measuring progress](/guide/#measurement) explains how to read results over time. Keep in mind that no tool can get you into an AI answer. Google says there are "no additional technical requirements" to appear in AI Overviews or AI Mode, so any tool that promises inclusion is overselling. ## Red flags in "most intuitive" lists 1. **No method.** The word "intuitive" appears, but no task or tester is described. 2. **A conflict of interest.** The publisher sells the winning tool or earns commission from it. 3. **The same praise for every entry**, which tells you nothing about how they differ. 4. **The vendor's own claims presented as findings.** 5. **No trade-offs.** Nothing about who the tool does not suit. A list with these flags may still be right. You just can't check it, so give it less weight than your own test. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you are trialling trackers, it is one option for checking whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions; run the same task list on it as on any other tool. ## Questions people ask ### Which AI search optimisation tool is the easiest to use? We did not find an independent usability ranking, so we cannot name one honestly. Run the same task list on two or three tools and let your own scores decide. ### Is a tool that is easy to use also accurate? Not necessarily. A clean interface tells you nothing about how the tool collects answers or how many runs sit behind each number, so check accuracy as a separate step. ### Can a small business owner use these tools without SEO knowledge? Some may suit you and some will not, and the quickest way to find out is to open the first report with no help. If you can explain what it says in two sentences, it works for you. ### Do I need a tool at all, or can I check AI visibility by hand? You can check by hand: ask your customers' questions, run each one more than once and log the answers in a spreadsheet. It costs nothing but time, and a tool earns its place once you have too many prompts to run yourself. ### How long should a trial take to judge a tool? No standard exists. We suggest about an hour of structured testing per tool as a starting point, with more time on the tasks where two tools came out close. ## Sources - [Semrush Knowledge Base: AI Visibility Toolkit](https://www.semrush.com/kb/1493-ai-toolkit) - [Ahrefs: Brand Radar](https://ahrefs.com/brand-radar) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) --- # Guide: Who is the world's most renowned AI visibility expert? The LinkedIn stunt, and what it shows about AI answers Canonical: https://whatisgeo.app/worlds-most-renowned-ai-visibility-expert/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-27 > Nobody holds an official title like this. It began as a consultant's tongue-in-cheek LinkedIn post that Google's AI Overview reportedly repeated, which shows how easily AI answers echo one confident source. No award, register or independent ranking stands behind the title. As far as we can trace, it started as a joke, so the more useful question is how to tell real expertise from one well-placed sentence. This page keeps three kinds of claim apart: what the people involved reported, what Google and published research document, and what we recommend. ## Is there an official world's most renowned AI visibility expert? No. We found no recognised certification, licence or objective ranking of AI visibility experts in the sources we checked. Course providers and vendor academies do issue their own certificates (see [AI SEO courses](/ai-seo-course/)), but a certificate from a seller is not an independent standard. If nobody can award a title, anybody can claim it. That says nothing about the claimant's skills either way. It means the title carries no weight of its own, so you need other evidence. ## Where the phrase came from According to a write-up by Sarah Schmidt on the PRson of Interest Substack (28 July 2026), Pedro Dias, whom it describes as "a former Google Search Quality Strategist turned AI visibility consultant", posted this on LinkedIn about a month earlier: > "OK guys... Just informing everyone I'm the world's most renowned AI visibility expert. Thank you for your attention to this matter." The write-up says that was the whole post and calls it tongue in cheek. It says Dias then searched Google for the phrase, and that the AI Overview cited the LinkedIn post as its primary source and stated the claim as fact, shown in a screenshot from Dias. Posts on X fill in the timing: - Dias posted on X on 27 June 2026: "Look. It's me. The worlds most renowned AI visibility expert". - Edward Sturm wrote on X on 28 June 2026 that Dias shared the AI Overview screenshot the same day as the LinkedIn post. - Replying to Sturm, Lily Ray wrote: "I tried the same day and didn't see it anywhere", and wondered how much was personalisation. - The Substack author reports searching again on 28 July and finding the AI Overview now described the phrase as "a tongue-in-cheek self-applied title and social media loop popularized on LinkedIn", with Dias still the first person cited. All of this is secondhand: posts and screenshots from the people involved. We have not reproduced it, and at least one person who tried the same day reported not seeing it. ## Why an AI answer could repeat it Google's documentation explains the general mechanics. Its Search Central page, AI features and your website, says AI Overviews and AI Mode may use "query fan-out", running several related searches to build a response, and then show supporting links. To be shown as a supporting link, a page must be indexed and eligible to appear in Google Search with a snippet, and Google says "there are no additional technical requirements." It also says AI Overviews only appear when its systems judge them additive to classic Search. The page does not say how Google weighs a self-declared claim against other sources. Our interpretation, which Google does not document: "world's most renowned AI visibility expert" is an unusual exact phrase. Very few pages are likely to use those words, so the page that coined them is the obvious match. A system summarising the best-matching sources can then end up repeating that page's claim. A common question with many competing sources is a different situation. Research supports a narrower point: the wording of a source affects how much of it appears in generated answers. Aggarwal et al., GEO: Generative Engine Optimization (accepted to KDD 2024), rewrote source pages and measured how visible they became in answers. Their main test engine took the top five Google results for each query and had GPT-3.5-turbo write the answer, across a benchmark of 10,000 queries, with a smaller test on Perplexity. Adding quotations, statistics and citations did best, with visibility gains of up to 40% on the paper's own metrics. The authors note that results varied by domain and that methods may need to change as engines evolve. The study measured the visibility of rewritten pages. It did not test whether false or self-declared claims get repeated. For how AI answers are put together more generally, see [how AI answers happen](/guide/#how-it-works). ## What the episode does and does not prove If the screenshots are accurate, a single confident post on a phrase nobody else used was repeated in an AI Overview and cited as the source. That proves nothing about whether the claim is true, or about Dias's skills. It does not show that AI Overviews behave this way on ordinary queries, how often it happens, or for how long. One reported case is an anecdote. Lily Ray's same-day failure to see it, and the reworded Overview the Substack describes a month later, suggest the answer changed over time and may have differed between users. Google's AI Overviews can also simply be wrong; we cover the evidence in [are Google AI Overviews accurate](/are-google-ai-overviews-accurate/). Treat a confident answer as a pointer to its cited sources, and check them. ## Why "top AI visibility experts" lists are weak evidence Search for this topic and you will find listicles ranking AI visibility experts. We will not repeat their rankings. What matters is how they were made. Take one we read: a post titled "Top AI Visibility Experts Dominating Search in 2026" on Indie Hackers (6 May 2026). Its entire stated method is that the author analysed "the online presence, reputation, and work of over 50 industry experts". It gives no criteria, no scoring, no list of who was considered and no way to check the order. Nothing on the page would let a reader reproduce the ranking. When you meet a list like this, look for: | Check | Why it matters | | --- | --- | | Stated criteria and weighting | Without them the order is opinion | | Who was considered and who was left out | Shows whether the field was actually surveyed | | Who wrote it, and any link to the people ranked | Self-interest changes how much weight it deserves | | A date and a way to verify | Lets you check the claims still hold | A list that fails all four is an opinion in numbered form. It is also exactly the sort of page an AI answer might pick up and repeat. ## How to vet someone who claims AI visibility expertise This is our recommendation, not an industry standard. With no independent certification to lean on, you vet by evidence. Ask for: 1. **Their method.** Can they explain, step by step, how they measure whether AI answers name and cite a business? Compare it with the approach in [how to measure AI visibility](/how-to-measure-ai-visibility/). 2. **Client results you can verify.** A named client you can contact, or a result you can see in public. "A major brand" is not checkable. 3. **Reproducible prompt tests.** Which prompts, which AI systems, on what dates, and what came back. Answers can vary between runs and users, so one screenshot proves little. 4. **A clear split between claims.** Good practitioners separate what a platform documents, what a named study found under its conditions, and what is their own judgement. If everything arrives at the same level of confidence, be wary. 5. **Who else says so.** If the only source calling someone an expert is that person, it is a self-description, however many times an AI repeats it. If you are hiring a firm, our guide to [choosing a GEO agency](/generative-engine-optimization-agency/) covers what to ask before signing. ## What to do if you want AI to describe you accurately Skip the stunt. Being named, being cited with a link and being described accurately are three separate outcomes, and a one-line claim does not reliably deliver any of them. Nobody can guarantee how an AI answer will describe you. What we recommend instead: - **Check what answers say now.** An [AI visibility audit](/ai-visibility-audit/) records which questions name you, what the answers say and which sources they cite. - **Cover Google's basics.** Google's guidance for AI features is ordinary Search practice: crawlable, indexed pages, important content available as text, structured data that matches the visible page, and up-to-date Business Profile information. - **Fix the sources answers rely on.** If an answer cites an outdated directory listing or profile, correct that page where you can. - **Back claims with evidence.** A claim other people can check holds up better than a title you gave yourself. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want to see what AI answers say about you rather than trust a screenshot, it checks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for your buyers' questions. - [Hilyt](https://hilyt.it/): If you want AI systems to find an accurate description of you or your business, it publishes an open, AI-readable profile with your facts in one place. ## Questions people ask ### Who is the world's most renowned AI visibility expert? Nobody holds that title. The phrase comes from a tongue-in-cheek LinkedIn post by consultant Pedro Dias, as reported by the PRson of Interest Substack, and we found no objective ranking of AI visibility experts in the sources we checked. ### Can you get an AI to call you an expert? Screenshots shared by the people involved suggest a Google AI Overview repeated a self-declared title on an unusual phrase, but we have not reproduced it and one tester reported not seeing it the same day. Nobody can guarantee what an AI answer will say, and a repeated claim is not proof of expertise. ### Does a LinkedIn post influence what Google AI Overviews say? It can be cited if it is indexed and eligible to appear in Search with a snippet, which Google says is the only technical requirement for supporting links. One reported case does not show how often posts shape answers on ordinary queries. ### How do you check whether an AI visibility expert is credible? We recommend asking for their method, client results you can verify, reproducible prompt tests, and a clear split between documented, researched and guessed claims. That is our recommendation, not an industry standard. ### Is there a ranking or certification for AI visibility experts? We found no recognised independent certification, licence or objective ranking in the sources we checked. Course providers issue their own certificates, and 'top experts' lists we read gave no reproducible method. ## Sources - [PRson of Interest (Sarah Schmidt), 28 July 2026: A Guy on LinkedIn Declared Himself the "World's Most Renowned AI Visibility Expert." Google Agreed.](https://prsonofinterest.substack.com/p/a-guy-on-linkedin-declared-himself) - [Pedro Dias on X, 27 June 2026](https://x.com/pedrodias/status/2070821596458057974) - [Edward Sturm on X, 28 June 2026 (with Lily Ray's reply)](https://x.com/edwardeachday/status/2071214234994671956) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Indie Hackers, 6 May 2026: Top AI Visibility Experts Dominating Search in 2026 (example of a list with no reproducible method)](https://www.indiehackers.com/post/top-ai-visibility-experts-dominating-search-in-2026-337caaab17) --- # Guide: Will AI replace SEO? Canonical: https://whatisgeo.app/will-ai-replace-seo/ Publisher: Adapt Progress Evolve Limited Updated: 2026-10-18 > No platform documentation or research we have read says AI search removes the need for SEO. Google says SEO basics still apply to its AI features. What changes is where you appear and how you measure it. Nothing we have read from Google, OpenAI, Perplexity or Anthropic, or in the best-known research paper on the topic, says that pages crawlers can reach stop mattering once AI writes the answers. The documented changes are narrower: where your content appears, and how you can see it. Anything about the longer term is a forecast. Most "SEO is dead" posts mix three kinds of claim in one paragraph. On this page each one is labelled: **documented** (a platform's own documentation says it), **research** (a named study, with its conditions and limits) and **our recommendation** (sensible practice, nothing more). Where a claim is only a prediction, we say so. ## Two different questions: the channel and the job "Will AI replace SEO?" asks two things at once. - **The channel question:** what happens to search when AI answers sit above or beside the usual results? - **The job question:** which of the tasks an SEO does can a tool now do? They have different answers. A channel can change shape while the work inside it stays busy, and a task can be partly automated while the channel grows. Treating them as one question is how a headline manages to sound alarming and tell you nothing. For the wider comparison of classic search work and AI answer work, see [GEO vs SEO](/guide/#geo-vs-seo). ## What platforms document about AI answers and ordinary search **Documented (Google).** Google's page on AI features and your website says that to be eligible as a supporting link in AI Overviews or AI Mode, "a page must be indexed and eligible to be shown in Google Search with a snippet", and that "there are no additional technical requirements." It adds that "all existing SEO fundamentals continue to be worthwhile". Its examples include allowing crawling in robots.txt and by any CDN or hosting infrastructure, internal links, important content available as text, and structured data that matches the visible page. Google's Search Central blog post on succeeding in its AI experiences (May 2025) says the same in plainer words: "The underpinnings of what Google has long advised carries across to these new experiences," and "Meeting the technical requirements covers you for search generally, including AI formats." **Documented (other assistants).** OpenAI, Perplexity and Anthropic each publish crawler documentation. It establishes one narrow point: access is a dependency. | Platform | Crawler used for search answers | What its documentation says about it | |---|---|---| | OpenAI (ChatGPT) | OAI-SearchBot (GPTBot is separate, for training) | Sites opted out of OAI-SearchBot "will not be shown in ChatGPT search answers, though can still appear as navigational links" | | Perplexity | PerplexityBot | Used to "surface and link websites in search results"; Perplexity recommends allowing it in robots.txt | | Anthropic (Claude) | Claude-SearchBot (ClaudeBot is separate, for training) | Disabling it "may reduce your site's visibility and accuracy in user search results" | None of these pages says the crawler will choose your page, rank it or send you visitors. What they show is that a page the bot cannot reach is out of the running before quality comes into it. OpenAI also documents that its search and training settings are independent, so you can allow one and block the other. More on this under [crawler access](/guide/#crawler-access). ## What research says about visibility in AI answers **Research.** The most cited paper is Aggarwal et al., "GEO: Generative Engine Optimization", accepted to KDD 2024. Its test conditions matter more than its headline: - The authors built their own generative engine. For each query it fetched the top five Google results, and a language model wrote an answer from them. - They tested on GEO-bench, a benchmark of 10,000 queries they put together from nine sources. - Visibility was scored with the authors' own metrics: how much of the answer drew on a source and how prominently, plus a subjective score rated by GPT-3.5. - They report that rewriting content to add citations, quotations and statistics raised visibility on those metrics, and that results varied by subject area. They also ran the methods on Perplexity.ai. The limits: this is one engine design, built for the study and measured with metrics the authors defined. It did not test ChatGPT or Google's AI features, and those products have changed since. It does not address whether SEO is still needed. One detail of the method is worth noticing. In that setup a page could only be cited if it was already in Google's top five for the query. Ordinary retrieval came first, and presentation was tested on top of it. ## What is genuinely changing for SEO Two things are documented: where visibility shows up, and how you measure it. **Where it shows up.** Google says AI Overviews and AI Mode may use a "query fan-out" technique, running several related searches to build a response, and can display "a wider and more diverse set of helpful links" than a classic results page. It also says AI Overviews appear only when its systems judge them additive to classic Search, "and as such, often don't trigger." So a page can be used inside an answer, shown as a supporting link, or left out entirely, and each of those looks different in a report. **How you measure it.** Google documents that sites appearing in AI Overviews and AI Mode are counted in the overall search traffic in the Search Console Performance report, under the "Web" search type. The page we read describes no separate AI-only report, so what you see is the combined figure. Google also says clicks from results pages with AI Overviews are "higher quality". That is Google's own observation, and the page publishes no figures behind it. For ChatGPT, Perplexity and Claude, none of the crawler pages we read offers a reporting tool. The [measurement](/guide/#measurement) section covers practical ways to check. On the platforms' own account, the dependency on pages that are reachable, indexed and eligible is unchanged. ## Which SEO tasks AI tools can take on, and which still need a person This is the job question. We have no study to cite here, so everything in this section is **our recommendation**. Treat AI tools as aids for drafting and analysis, and have a person review what they produce. They are quick at first drafts of routine copy, sorting and summarising lists of pages or queries, and spotting patterns in data you have already collected. A person still needs to make the judgement calls. Is this page factually right? Does it answer the question a buyer actually has? Should it be published at all? A tool produces fluent text whether or not its claims hold up, and fluent text can still be wrong. We will not put a number on how much of an SEO's work this covers, because we have found no source that measures it. ## What to keep doing, and what to start checking **Keep doing:** 1. **Keep pages crawlable and indexable.** Google documents this as the requirement for AI Overviews and AI Mode, including allowing crawling at the CDN or hosting level. 2. **Make sure structured data matches what is visible on the page.** Google lists this among the SEO fundamentals for AI features. 3. **Keep factual pages accurate and clear.** Our recommendation: if a machine summarises you, you want the summary to be right. 4. **Decide, crawler by crawler, what your robots.txt allows.** OpenAI and Anthropic document separate crawlers for search and for training, so this is a real choice. **Start checking (our recommendation):** 1. Which AI assistants name your business when someone asks a buyer question in your market. Ask the questions yourself on more than one assistant, and record the answers over time. 2. Whether the documented crawlers can actually reach the pages that matter. Your server logs show which user agents visit. 3. What Search Console can and cannot tell you, given that AI feature traffic is reported inside the overall Web figures. None of this comes with a timeline or a guarantee, and nobody can promise inclusion in an AI answer. For background on how answers are put together, see [how AI answers happen](/guide/#how-it-works). For where AI fits into SEO work more generally, see [what AI SEO means](/what-is-ai-seo/). ## How to judge the next "SEO is dead" headline Ask four questions: 1. Does it name a source? 2. Is that source platform documentation, a piece of research, or an opinion? 3. If it is research, does it give the conditions the result holds under? 4. Does it separate the channel from the job? Most posts fail the first question. Some pass it and fail the third, quoting a study without its limits. Even the GEO paper's abstract says generative engines are "rapidly replacing traditional search engines like Google and Bing". That is the authors' framing: their experiments measure visibility inside answers and do not measure how many people use search. A confident claim with no source behind it is an opinion, however it is presented. Google's own blog post makes the only forecast we would repeat: "The only thing predictable in Search is that it always evolves." ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want to see whether ChatGPT, Claude, Gemini and Perplexity actually name and cite your company for your buyers' questions, it does that check for you. ## Questions people ask ### Is SEO going away with AI? No source we have read says so. Google documents that pages need to be indexed and eligible to show with a snippet to appear in AI Overviews and AI Mode, and that existing SEO fundamentals continue to be worthwhile. ### Is SEO dead after ChatGPT? We found nothing that shows it. OpenAI documents that sites which block its search crawler, OAI-SearchBot, will not be shown in ChatGPT search answers. That makes crawler access a dependency, which is classic SEO ground. ### Will ChatGPT replace SEO? Nobody can say, and any answer is a forecast. ChatGPT's own documentation describes relying on a crawler to surface websites in its search answers, and the best-known research paper on the topic did not test ChatGPT at all. ### How will ChatGPT affect SEO? The documented effects are narrow: your content can appear inside answers as well as in result lists, and that is harder to measure. Our recommendation is to keep pages crawlable and accurate and to check regularly which assistants name your business for buyer questions. ### Will SEO exist in five years? Nobody can predict that, and a source that claims to know is guessing. What we can tell you is what platforms document today, and today they all depend on pages their crawlers can reach. ## Sources - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [Google Search Central Blog: Top ways to ensure your content performs well in Google's AI experiences on Search (May 2025)](https://developers.google.com/search/blog/2025/05/succeeding-in-ai-search) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Perplexity: Perplexity crawlers](https://docs.perplexity.ai/guides/bots) - [Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) - [Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)](https://arxiv.org/abs/2311.09735) - [Aggarwal et al., GEO: Generative Engine Optimization (full text, v3)](https://arxiv.org/html/2311.09735v3) --- # Guide: Yoast SEO AI: what its AI features do, what they cannot do, and what to check first Canonical: https://whatisgeo.app/yoast-seo-ai/ Publisher: Adapt Progress Evolve Limited Updated: 2026-11-10 > Yoast's AI features draft titles and meta descriptions, control AI crawler access and llms.txt, and in the AI+ plan report how AI assistants mention your brand. None of them can guarantee you appear in an AI answer. Yoast's AI features do three different jobs. Some help you write: titles, meta descriptions and summaries. Some control what AI crawlers and agents can read on your site: the bot blocker, llms.txt and schema output. One, in the separate Yoast SEO AI+ plan, reports how AI assistants mention your brand. Everything below about the products comes from Yoast's own product, feature and release pages and its WordPress.org listing, read on 30 October 2026. Where we give advice, we say it is ours. ## What Yoast SEO's AI features are | Feature | What Yoast says it does | Where Yoast lists it | Job | |---|---|---|---| | AI Generate | Drafts titles and meta descriptions, including social titles, from what you have written | Premium, WooCommerce SEO, AI+ and Yoast SEO for Shopify | Writing | | AI Optimize, AI Summarize, AI Content Planner | Suggests keyphrase edits, writes a post summary, suggests post ideas and a starter draft | Premium (the listing says Yoast's AI tools "come with Yoast SEO Premium") | Writing | | Bot blocker | Adds robots.txt lines that block GPTBot, CCBot and Google-Extended | Premium | Crawler access | | llms.txt | Generates and refreshes an llms.txt file | Free plugin and every paid plan | File generation | | Schema.org output | Structured data; Premium adds "significantly more schema types" and schema aggregation for NLWeb | Free, with more in Premium | Markup | | AI Brand Insights | Weekly report on how ChatGPT, Gemini, Claude and Perplexity mention your brand | Yoast SEO AI+ | Measurement | On 30 October 2026 Yoast's product pages showed Yoast SEO Premium at $118.80 a year and Yoast SEO AI+ at $358.80 a year, both excluding VAT. The AI+ page says it includes Yoast WooCommerce SEO and the Premium features. Yoast's pages have a currency switcher (USD, EUR or GBP). Read from Europe on 10 November 2026, the Premium and AI+ pages showed the same figures in euros, €118.80 and €358.80 a year ex VAT, and the AI+ page's structured data lists the price as USD 358.80. Prices change, so check on the day you buy. The WordPress.org listing says "All AI tools included, no hidden fees." Read that in context. The same listing says the AI tools come with Premium, with "no separate AI subscription, no extra login, and no credit top-up to buy". So you pay nothing on top of a paid plan, but the AI writing tools themselves are Premium features. None of these is a way into AI answers. Each section below says what the feature does and where its job stops. ## The writing helpers: AI-generated titles and meta descriptions According to the WordPress.org listing, Yoast AI Generate "creates five SEO-friendly titles and meta descriptions", with one-click regeneration if you want more options. Yoast's feature page adds that it works for posts, pages and category pages and can draft social sharing titles and descriptions. You apply or dismiss each suggestion. The bulk editor is free; drafting metadata in bulk with AI needs Premium. What you get is a first draft of a snippet. You still choose, edit and publish. Our recommendation is to check every generated line against the page it describes. Does the title say what the page actually covers? Does the description promise something the content does not deliver? Generated copy tends to sound plausible even when it is slightly off, and that is how errors slip through. A better meta description helps people scanning search results; that is the whole of its job. Google's documentation on AI features says a page needs to be indexed and eligible to show in Google Search with a snippet to appear as a link in AI Overviews or AI Mode, and that "there are no additional technical requirements." Nothing in that documentation makes a generated meta description a route into those answers. ## Blocking or allowing AI crawlers from Yoast Yoast SEO 21.5, announced on 31 October 2023, added a Premium toggle to block AI bots. Yoast's release post says it blocks "AI robots like GPTBot, CCBot, and Google-Extended" by automatically adding disallow lines to your robots.txt file. The current listing names the same three. What those tokens control is set by the platforms, not by Yoast: - **GPTBot:** OpenAI says disallowing it indicates your content should not be used to train its foundation models. ChatGPT search uses a separate crawler, OAI-SearchBot, and OpenAI says each setting is independent of the others. - **Google-Extended:** Google says it controls whether crawled content may be used to train future Gemini models and for grounding in Gemini Apps and Vertex AI, and that it "does not impact a site's inclusion in Google Search." So the toggle, as Yoast describes it, is about AI training. Yoast's own heading for the feature is "Blocking bots used to train AI systems". Its post and listing do not name OAI-SearchBot, PerplexityBot or ClaudeBot, so do not assume it covers them. Our guide to [AI crawlers and robots.txt](/ai-crawlers-robots-txt/) lists the documented tokens for each platform. Two more limits. First, robots.txt is a request to crawlers: the standard, RFC 9309, says its rules "are not a form of access authorization." Second, blocking or allowing is your decision, as Yoast's post says under "Think before you do". Our recommendation: decide what you want first, flip the toggle, then open yoursite.com/robots.txt and read exactly which lines were added. The home guide's section on [crawler access](/guide/#crawler-access) covers the trade-offs. ## llms.txt and structured data in Yoast: what is documented **llms.txt.** Yes, Yoast generates one. Its feature page says you can generate the file "automatically with a click" and that it refreshes automatically. Yoast can pick your key content or you can choose it manually, and you can preview the file before publishing. The page calls it "free for everyone" and lists the free plugin, Premium, WooCommerce SEO, AI+ and Yoast SEO for Shopify. What Yoast cannot tell you is whether AI tools will read the file. llms.txt is a proposal published at llmstxt.org, not a standard, and no assistant has to use it. Yoast's page says the file can help AI tools represent your business more accurately; that is Yoast's description, not a commitment from any AI platform. Our [llms.txt guide](/llms-txt/) covers what is and is not known. **Structured data.** The listing says the free plugin outputs Schema.org structured data. It says Premium "outputs significantly more schema types" and adds schema aggregation for NLWeb, which combines your site's structured data into one graph. For developers, Yoast's Schema API covers graph pieces including Article, Organization, Person, Breadcrumb and WebPage. Its HowTo and FAQ blocks output schema by default. The listing does not say which types appear on which of your templates. To see what your pages actually output, test them yourself. Google's AI features documentation asks that structured data match the visible text on the page, and repeats that there are no additional technical requirements for AI features. Our guide to [schema markup for AI](/schema-markup-for-ai/) covers what structured data can and cannot do. ## AI brand tracking: what Yoast reports and how to treat the numbers Yoast SEO AI+ includes Yoast AI Brand Insights. According to Yoast's AI+ page: - You set up a brand profile, choose competitors and create up to five brand questions, or let AI suggest them. - Each brand gets one scheduled analysis a week across ChatGPT, Gemini, Claude and Perplexity. Yoast says no extra automatic analyses are available. - The dashboard shows an "AI visibility index": a score out of 100 that blends mentions, ranking, citations and sentiment. It also shows Found or Not found for each question, the sources cited, a competitor ranking and a positive sentiment percentage. - It does not need access to your website's admin. How we suggest reading those numbers: five questions checked once a week is a small sample. AI answers can vary between runs, users and platforms, and the AI+ page we read does not say how many times each question is run. A blended score out of 100 also hides which part of it moved. So treat the dashboard as a lead, then check it. If it says you were cited, look at the actual answer. If sentiment drops, read the wording behind it. Keep an independent check running alongside: ask the same questions yourself on a schedule and record what comes back. Our guide on [how to measure AI visibility](/how-to-measure-ai-visibility/) sets out a method, and the [AI visibility checker](/ai-visibility-checker/) page covers quicker spot checks. ## What Yoast cannot do for AI visibility Yoast says it plainly in the FAQ on its own listing: "Nobody can guarantee that an AI assistant will mention your site." We agree, and we recommend treating any plugin the same way. Yoast can draft snippets, write robots.txt lines, generate an llms.txt file, output schema and report on mentions. None of those makes an AI system name or cite you. Some things stay outside any plugin's reach: the quality of your content, whether other sources reference you, and whether an AI system chooses to retrieve your page at all. ## How to decide what you need, and how to check it yourself Our recommended order: 1. **Name the job.** Do you want writing help, crawler control or measurement? Pick one before you look at plans. 2. **Check what is free.** According to Yoast's pages, llms.txt generation and basic schema are in the free plugin. The AI writing tools and the bot blocker are Premium. Brand Insights is in AI+ only. 3. **Check the price on the day.** Use Yoast's own product pages and note the date. 4. **For crawler control,** confirm which user agents the toggle adds, then read your live robots.txt. 5. **For llms.txt and schema,** open the generated file and test a few pages after you switch them on. 6. **For measurement,** run your own checks alongside any dashboard before you act on a number. 7. **Set a review date.** Yoast's listing says the plugin is updated every two weeks, so features and plans change. ## From the same studio These are made by the studio that publishes this guide. - [visibilitytoai](https://visibilitytoai.com/): If you want a second, independent reading to compare with Brand Insights, it checks whether ChatGPT, Claude, Gemini and Perplexity name and cite your company for buyers' questions. ## Questions people ask ### Is Yoast SEO AI free, or do I need Premium? Partly free. Yoast's pages put llms.txt generation and basic schema in the free plugin. AI Generate, the other AI writing tools and the AI bot blocker are Premium features. On 30 October 2026 Premium was $118.80 a year excluding VAT. ### Does Yoast SEO AI help me show up in ChatGPT or Google AI Overviews? No Yoast feature can guarantee that, and Yoast's own WordPress.org listing says nobody can. Google says AI Overviews and AI Mode have no technical requirements beyond normal Search eligibility, so the basics of your site matter more than any plugin setting. ### Can Yoast block AI bots from my WordPress site? Yes, in Premium. Since version 21.5 a toggle adds robots.txt rules for GPTBot, CCBot and Google-Extended, which Yoast describes as bots used to train AI systems. It does not name search crawlers such as OAI-SearchBot, so read your robots.txt afterwards. ### Does Yoast generate an llms.txt file? Yes. Yoast's feature page says the free plugin and all paid plans generate and refresh an llms.txt file automatically, and you can choose the pages yourself. Whether an AI assistant reads the file is up to that assistant. ### Is Yoast SEO AI+ worth it, and how do I check its brand tracking numbers? That depends on whether you need weekly brand monitoring. On 30 October 2026 AI+ cost $358.80 a year excluding VAT and covered up to five questions per brand, analysed weekly across ChatGPT, Gemini, Claude and Perplexity. To check its findings, ask the same questions yourself and read the actual answers. ## Sources - [WordPress.org: Yoast SEO plugin listing (read 30 October 2026)](https://wordpress.org/plugins/wordpress-seo/) - [Yoast: Yoast SEO 21.5: Helping you block AI bots in Premium (31 October 2023)](https://yoast.com/yoast-seo-october-31-2023/) - [Yoast: AI Generate feature page](https://yoast.com/features/ai-generate/) - [Yoast: llms.txt feature page](https://yoast.com/features/llms-txt/) - [Yoast: Yoast SEO AI+ product page (price read 30 October 2026)](https://yoast.com/yoast-seo-ai-plus/) - [Yoast: Yoast SEO Premium product page (price read 30 October 2026)](https://yoast.com/product/yoast-seo-premium-wordpress/) - [OpenAI: Overview of OpenAI crawlers](https://platform.openai.com/docs/bots) - [Google: Google's common crawlers (Google-Extended)](https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers) - [Google Search Central: AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) - [RFC 9309: Robots Exclusion Protocol](https://www.rfc-editor.org/rfc/rfc9309) - [llmstxt.org: The /llms.txt file proposal](https://llmstxt.org/) ## AI agents - [MCP server](https://whatisgeo.app/mcp): Read-only remote MCP server (Streamable HTTP, no key). Tools: search_site, get_page, list_pages. Covers every page in the sitemap, with its URL and last-updated date. - [Connect the MCP server](https://whatisgeo.app/mcp-server/): How to add it to Claude, ChatGPT, Claude Code, Cursor or VS Code, what each tool does and what we record. - [MCP discovery card](https://whatisgeo.app/.well-known/mcp.json): The server's endpoint, transport and tools as JSON.