What GEO covers
A generative engine writes a new response instead of returning a ranked list. It may draw on its training, on pages it fetches at the time of the question, or on both. What a given tool does varies, and the providers describe their systems only in outline.
GEO is the practical response to that uncertainty. You cannot see inside the model, but you can control what your own site says, whether crawlers can read it, and how consistent your business details are across the web. Those are the same raw materials every tool has to work with.
The term overlaps heavily with answer engine optimisation. We use GEO when the question is about the whole picture of your business, such as whether an assistant gets your services, area and contact details right. We use AEO when the question is about one page answering one query. Google treats both terms as SEO: its guide says optimising for generative AI search is optimising for the search experience.
What the documentation says
We rely on what providers publish rather than on folklore. This is what we found when we rechecked on 23 October 2026.
| Provider | What is documented | What is not documented |
|---|---|---|
| AI features use Search's indexing and snippet eligibility, on sites not excluded in Search Console; no special markup or AI files are needed | Exactly how links are chosen inside an AI response | |
| OpenAI | OAI-SearchBot surfaces sites in ChatGPT search; GPTBot relates to training; ChatGPT-User acts on a user's request; OAI-AdsBot checks pages submitted as ads | How ChatGPT ranks the sources it shows |
| Anthropic | Three named bots (training, user requests, search) each have a stated purpose and honour robots.txt | How Claude selects pages for an answer |
| Perplexity | PerplexityBot surfaces sites in results and is not used for training; Perplexity-User fetches pages for user requests | How results are ordered or chosen |
| Microsoft | Bing Webmaster Tools reports citations in Copilot and Bing AI summaries | How placement within an answer is decided |
Entity consistency
An entity is simply a thing a system can identify: your business, a service, a suburb. Tools find it easier to describe you correctly when your name, services, service area and contact methods read the same across your website, any directory you control and your Google Business Profile.
Bing's own guidance for publishers makes a similar point: keep text, images and video consistent about the same entities and concepts. For place-based businesses, Bing also points to Bing Places for Business as the way to keep address, hours and contact details current. We audit those details line by line and list the conflicts, because contradictions are common and easy to fix.
Where structured data helps and where it stops
Structured data is useful, but its limits are often overstated in AI search talk.
- It must match the page: Google says markup should reflect the visible text, so we never mark up claims the page does not make.
- It is not required for Google AI features: Google states no special schema.org markup is needed for AI Overviews or AI Mode.
- It still supports rich results: Google recommends continuing to use it where it makes pages eligible for rich results.
- It does not fix weak content: Markup describes a page; it does not make a thin or vague page worth citing.
What we do, step by step
A typical GEO engagement runs through these stages, in this order.
- Baseline testWe ask a fixed list of real customer questions across several AI tools and record how, and whether, your business appears.
- Access reviewWe check robots.txt, meta robots tags and CDN or firewall rules against each provider's published crawler list.
- Fact auditWe compare your details across your site and public profiles and list every conflict or gap.
- Page improvementsWe strengthen the pages that matter, adding first-hand detail that a general summary could not supply.
- RetestWe repeat the baseline questions and report differences, with the caveat that responses vary between runs.
Tactics we avoid
Some popular advice either has no support in the documentation or carries risk.
What we do
- Add first-hand detail, such as how your process really works.
- Fix contradictions in business details.
- Keep the crawlers you want to be seen by allowed.
- Test and report with dates and wording.
What we do not do
- Create a page for every imagined AI sub-question, which Google says risks its scaled content abuse policy.
- Buy or fake mentions; Google says inauthentic mentions will not help much.
- Claim a proprietary score for AI visibility.
- Promise a result.
Questions to put to any GEO provider
Google's guidance on third-party advice says good advice either labels its claims as opinion or cites official guidance, and that no outside tool can see Google's internal systems. Use these questions to test any offer, ours included.
| Ask | A sound answer | A warning sign |
|---|---|---|
| What will you change on my site? | A written list of pages, access rules and facts, with reasons | Vague talk of an AI-ready package |
| How do you measure results? | Dated tests with the exact wording, plus platform reports | A proprietary visibility score |
| What do you need from me? | Confirmed facts and access to your settings | Nothing, or payment before any review |
| Can you promise a mention? | No, because no provider publishes a selection method | Yes, or a guarantee of placement |
| Which official sources do you rely on? | Named documentation from Google, OpenAI, Anthropic, Perplexity and Microsoft | Unnamed studies or secret signals |
Where GEO fits with other services
GEO depends on the groundwork of technical SEO, particularly crawling and indexing. For businesses serving a local area, tidy listings under Google Business Profile support the entity work. If your main concern is a particular surface, see Google's AI features or ChatGPT. The full map of this topic is on the AI search page.
Frequently asked questions
Is GEO different from SEO?
It is an extension rather than a replacement. SEO gets pages crawled, indexed and ranked. GEO asks a further question: when an AI tool writes a response about your field, does it have accurate, quotable information about you? Much of the work is shared, particularly crawl access and clear content.
Can GEO make ChatGPT recommend my business?
We cannot promise that. OpenAI documents what its crawlers do but does not publish how ChatGPT chooses what to show. What we can do is make your site readable to the right crawler, correct inconsistent details and test whether your business appears for real questions.
Do I need an llms.txt file?
Not for Google: it states Search ignores llms.txt and similar AI files. The file is a proposed convention, and the crawler pages from the other three providers make no commitment to it. It is harmless to add, but we would not rely on it.
Should I block AI crawlers?
It is a business decision. Some crawlers collect training data, while others fetch pages to answer questions or surface your site in search. Blocking a search or retrieval crawler may reduce visibility in that product. We explain each option and leave the choice to you.
How do you know what AI tools say about my business?
We ask them. Using a fixed list of customer-style questions, we record the answers, the sources shown and the date. It is manual and imperfect, because answers vary, and the record shows the wording tools use and where they get facts wrong.
Will schema markup get me cited?
Not on its own. Google states no special markup is required for its AI features, and Bing and others describe clear content and consistent entities as the focus. Use markup to describe your page accurately, not as a lever for citations.
Sources and further reading
- Google Search Central: AI features and your website: No special markup or AI files needed; indexing and snippet eligibility.
- Google: Optimizing for generative AI features on Google Search: llms.txt not used by Google Search; scaled content abuse; inauthentic mentions; structured data; GEO is still SEO from Google's view.
- OpenAI: Overview of OpenAI crawlers: Purposes of OAI-SearchBot, GPTBot, ChatGPT-User and OAI-AdsBot.
- Anthropic: Does Anthropic crawl data from the web: Purposes of ClaudeBot, Claude-User and Claude-SearchBot; robots.txt support.
- Perplexity: Perplexity crawlers: Purposes of PerplexityBot and Perplexity-User.
- Microsoft Bing Blog: Introducing AI Performance in Bing Webmaster Tools: Published 10 February 2026 (page metadata). Citation reporting, publisher guidance on consistent entities and Bing Places for local details.
- llms.txt: A proposal to provide information to help LLMs use websites: llms.txt is a proposal, now at version 2; not a formal standard.
- Google Search Central: Guidance on using third-party SEO tools, services and advice: Third-party AEO and GEO tools cannot access Google's ranking data or guarantee performance.
- Search Console Help: Search generative AI control: Include is the default; Exclude removes a site from AI Overviews and AI Mode.