What Is GEO? What Actually Works for Visibility in AI Search (2026)
What GEO is and how it differs from SEO: sourced guidance on Google-Extended, llms.txt and AI bots, the steps that genuinely work and how to measure progress.
Short answer: GEO (generative engine optimisation) is the work of making sure a brand is described accurately, and cited as a source, in generative AI answers: ChatGPT, Gemini, Perplexity, Claude, and Google's AI Overviews and AI Mode. According to Google's own guidance, updated in July 2026, from Google Search's perspective this is still SEO: you do not need a special file, special schema or content "written for AI" to appear in AI answers. What makes the difference is a crawlable site, original and well-sourced content, and consistent company information everywhere it appears.
This article sets out what actually works, which claims are exaggerated, and how to measure progress, based on official documentation. The full source list is at the end.
Are GEO, AEO and SEO the same thing?
The three acronyms look at the same question from different angles: when people search, does your content show up?
- SEO: visibility in search engine results.
- AEO (answer engine optimisation): being the source for direct answers (featured snippets, voice assistants, AI answers).
- GEO (generative engine optimisation): being mentioned and linked in generative AI answers.
Google's guide "Optimizing your website for generative AI features on Google Search" is explicit: from Google Search's perspective, optimising for generative AI search is optimising for the search experience, and is therefore still SEO. Google's AI features draw on the same index and ranking systems; when an answer is generated, relevant pages are retrieved from the search index (grounding). The model may also split a question into several related sub-queries and fetch results for each (query fan-out).
So GEO is best treated not as a separate channel that replaces SEO, but as a measurement and content discipline built on the same foundations.
What does it take to appear in Google's AI answers?
According to Google's "AI features and your website", there are no additional requirements to appear in AI Overviews or AI Mode. A page needs to be:
- crawlable by Google (not blocked to Googlebot in robots.txt),
- indexed,
- eligible to be shown in Search with a snippet.
The controls are the familiar ones: nosnippet, data-nosnippet, max-snippet and noindex. Traffic from AI features is reported within overall search performance in Search Console, and Google also points site owners to a separate Generative AI performance report there.
What Google-Extended does and does not control
A common mistake is to treat Google-Extended as permission to appear in Google's AI answers. According to Google's crawler documentation, Google-Extended manages whether content Google crawls may be used to train future Gemini models and for grounding in Gemini Apps and Vertex AI. The documentation states plainly that Google-Extended does not affect a site's inclusion in Google Search and is not used as a ranking signal.
In practice:
- Blocking Google-Extended does not remove your site from AI Overviews or AI Mode; those are governed by Googlebot and snippet controls.
- Allowing Google-Extended does not make you appear in Google's AI answers.
- It is a content-use preference, and the decision belongs to your company's policy on training and reuse.
AI bots: search, user requests and training are different things
When you set bot permissions, separate each bot by purpose. There is no single "AI bot".
| Company | Search / index | User-initiated access | Model training |
|---|---|---|---|
| Googlebot (Search, including AI Overviews and AI Mode) | — | Google-Extended (training and Gemini grounding preference) | |
| OpenAI | OAI-SearchBot (ChatGPT search results) | ChatGPT-User | GPTBot |
| Anthropic | Claude-SearchBot | Claude-User | ClaudeBot |
| Perplexity | PerplexityBot | Perplexity-User | Perplexity says PerplexityBot is not used to train foundation models |
What this means in practice:
- According to OpenAI, blocking GPTBot does not affect appearing in ChatGPT search; to opt out of ChatGPT search results you block OAI-SearchBot.
- User-initiated fetchers (ChatGPT-User, Perplexity-User) may not apply robots.txt rules, because a person requested the page.
- Blocking training bots is a legitimate choice and does not, on its own, switch off search visibility.
What actually works
Google's guide and the other engines' documentation point the same way:
1. Original, non-commodity content. Google distinguishes between "7 tips" content that anyone could write and content carrying first-hand experience or expert judgement. AI answers draw on many sources; pages that repeat what others have said do not stand out. Publish your own data, field observations, decision tables and dated sources.
2. A readable structure. Paragraphs, meaningful headings, a short answer straight after the question and the detail afterwards. Write it for people; Google says there is no need to break content into tiny chunks or to write in a special style for AI.
3. A sound technical base. Crawlable pages, main content that is available without JavaScript, canonical URLs, clean redirects, and pages that load quickly and work well on mobile.
4. Consistent entity information. Company name, addresses, founding details, service scope and contact details should match across the website, Google Business Profile, LinkedIn and directories. Conflicting information can lead to an inaccurate description of your brand in AI answers.
5. Images and video. Google notes that its AI features can surface relevant images and video, and that following existing image and video SEO practice is enough.
6. Local and commerce data. Google Business Profile and Merchant Center data help services and products appear in both AI answers and classic results.
What is exaggerated or simply wrong
"Add llms.txt and AI will read you." llms.txt is a proposal made by Jeremy Howard in September 2024; it is not a binding standard, and there is no general commitment from engines to read it. Google states that Google Search does not use these files and that they will neither help nor harm visibility or rankings. Keeping one as an accurate, current summary for other tools is harmless; it should not be presented as a guarantee of visibility, crawling or citation.
"AI needs special schema." According to Google, structured data is not required for generative AI search and there is no special schema.org markup to add. Structured data is still useful for rich result eligibility and for clear entity information, provided it describes only what is visible and accurate on the page.
"FAQPage markup gets you an FAQ box." Google removed the FAQ rich result entirely in May 2026, and How-to rich results are no longer shown either. Genuine questions and answers on the page help readers, but do not expect the markup to produce a special appearance in Google.
"Create a page for every variation of the question." Google says that producing content for every query or fan-out variation, primarily to manipulate rankings, violates its scaled content abuse policy.
"Collect brand mentions." Google says that seeking inauthentic mentions is not as helpful as it might seem, and that its spam systems filter them out.
"This tool can see Google's AI data." According to Google, no third-party tool has access to its internal ranking or AI systems.
How to measure progress
AI visibility cannot be judged from a single answer; responses vary by person, location, language and time. A practical measurement set-up:
- Search Console: the overall performance report and the Generative AI performance report; impression and click trends for brand and service queries.
- GA4: the "AI Assistant" channel in the default channel group, sessions from referrers such as chatgpt.com, perplexity.ai and gemini.google.com, and how those sessions convert into enquiries, calls and WhatsApp messages.
- A reusable query set: keep a fixed list of brand, service and industry questions. For each check, record the engine, date, language, query, whether the brand was mentioned, how it was described and the source links given. Repeat the same set monthly and look at the trend. Never read a single answer as overall success or failure.
- Tracking inaccurate descriptions: if an engine describes your brand incorrectly, check that the correct information appears once, clearly, on your site and your external profiles.
A checklist for the first 30 days
- Are Googlebot and search bots blocked anywhere in robots.txt? Are your important pages indexed?
- Do important pages carry
nosnippetor an unnecessarymax-snippetrestriction? - Do company details (founding, address, phone, service scope) match across the site, Google Business Profile and LinkedIn?
- Do your service pages answer "what is it, who is it for, what is included, how does the process work, what drives the cost, how is success measured"?
- Do figures, research and claims show their source and date on the page?
- Has the company decided its position on training bots, and is that decision written correctly in robots.txt?
- Has the query set been created and the first measurement recorded?
How we approach this at PAN
In our SEO and GEO service we treat AI visibility not as a separate trick but as a measured outcome of technical SEO, content and consistent company information: first we measure crawling and indexing, then which questions the content genuinely answers, and finally mentions per engine using a query set. We do not guarantee rankings or inclusion in AI answers; each month we share in writing what we measured and what we changed. A summary of our approach is on our GEO page.
Sources
- Google Search Central, "Optimizing your website for generative AI features on Google Search", last updated 10 July 2026: https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- Google Search Central, "AI features and your website", last updated 10 December 2025: https://developers.google.com/search/docs/appearance/ai-features
- Google, "Google's common crawlers" (Google-Extended), last updated 14 July 2026: https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers
- Google Search Central, FAQ structured data and deprecation note: https://developers.google.com/search/docs/appearance/structured-data/faqpage
- OpenAI, "Overview of OpenAI Crawlers": https://developers.openai.com/api/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
- Perplexity, "Perplexity Crawlers": https://docs.perplexity.ai/guides/bots
- The llms.txt proposal, Jeremy Howard, 3 September 2024: https://llmstxt.org/
Sources checked on 4 October 2026. Because the rules in this area change often, we update this article as the documentation changes.