SEO SentinelBeta
Menu

Draft under editorial review. This page is not yet publication-ready and is excluded from search engines.

Emerging topic: needs an expert fact-check of the terminology and of the Google AI-features statements against current documentation before publication.

GEO & AI search

What is a GEO audit? AI-search readiness, explained

What a GEO (generative engine optimisation) audit can and cannot check: AI crawler access, content structure, entity consistency, sources and snippet controls.

By the SEO Sentinel team Updated 9 min read

GEO (generative engine optimisation) is the practice of making a website easy for AI-driven search and answer systems to access, understand and attribute. These are systems such as ChatGPT search, Perplexity, Claude, and Google’s AI Overviews and AI Mode. A GEO audit reviews the observable signals on a site that affect that, much as a technical SEO audit reviews crawlability and indexability for classic search.

The term comes partly from academic work on how generative engines select and cite sources. In agency practice it has come to mean a readiness review. That distinction matters, because the most important thing to understand about a GEO audit is what it can’t do.

What a GEO audit cannot tell you

A GEO audit looks at your site. It cannot observe what an AI system will say:

  • It can’t predict or guarantee mentions, citations or recommendations in AI answers.
  • It can’t measure “AI rankings”. Answers vary with the question wording, time, location, account, conversation history and model.
  • It can’t judge authority or factual accuracy in any meaningful way.

Any tool that turns a readiness review into a single “GEO score” or promises citations is claiming more than the evidence supports. To learn what AI tools actually say about a brand, you need observations: ask the questions and record the answers over time. That is a separate activity from a GEO audit.

Google states that its AI features in Search have no additional technical requirements beyond being eligible for normal Search (Google: AI features and your website). For Google, then, good technical SEO is most of the work. Other AI search providers run their own crawlers, which is where GEO-specific checks start.

The four areas of a GEO audit

1. AI crawler access

Can AI search crawlers fetch the pages? Check robots.txt rules for each documented crawler, grouped by purpose:

  • AI search crawlers (e.g. OAI-SearchBot, Claude-SearchBot, PerplexityBot): blocking them can keep a site out of those products’ search features.
  • Training crawlers (e.g. GPTBot, ClaudeBot): blocking them is a legitimate business choice, not an error.
  • Control tokens (Google-Extended, Applebot-Extended): robots.txt tokens with no separate crawler of their own.
  • User-triggered fetchers (e.g. ChatGPT-User): their operators say robots.txt may not apply.

Also remember that robots.txt permission doesn’t prove access. CDNs, web application firewalls and bot-protection services often block AI crawlers regardless of robots.txt. Server logs and the operators’ published verification methods are the evidence. Our AI crawler robots.txt guide covers the details.

2. Content readiness

Is the content in the HTML and easy to parse into self-contained passages?

  • Main content present in the server-rendered HTML. A page that’s an empty JavaScript shell gives non-rendering crawlers nothing to work with.
  • Descriptive headings rather than “Overview” or “More info”.
  • Long text broken up with subheadings; very long paragraphs split.
  • Question-style headings followed by an actual answer.
  • Snippet controls checked deliberately: nosnippet, max-snippet:0 and data-nosnippet restrict how content can be shown, and Google applies them to its AI features too (Google: robots meta tag).

Don’t flag a page for lacking an FAQ section. Not every page needs one, and inventing questions makes content worse.

3. Entity signals

Is it unambiguous who the site belongs to?

  • The organisation name matches across Organization structured data, og:site_name, page titles and the copyright line.
  • Organization markup exists, with a name, URL and logo.
  • About and Contact pages are linked and crawlable.
  • Contact details in structured data match what’s visible on the page, e.g. the same phone number as the tel: link.

4. Sources and evidence

On article content (pages marked up as Article, BlogPosting or NewsArticle):

  • a visible author and publication date, consistent with the structured data
  • consistent published and modified dates
  • outbound links to sources for factual claims
  • quotations with visible attribution

These are good editorial practice with or without AI search, and they make a passage easier to attribute.

How to present GEO findings to a client

  1. State the boundary first: “This reviews readiness. It does not measure whether AI tools mention you.”
  2. Separate decisions from defects. A blocked training crawler is a policy choice to confirm with the client; a blocked AI search crawler is usually a problem to discuss.
  3. Mark certainty. Something in robots.txt is confirmed; “no About page found” on a partial crawl is unverified.
  4. Group template-level findings. If the same content issue appears on most pages of a type, it’s one fix.
  5. Pair it with observations if the client wants to know about visibility: a fixed set of questions, asked and recorded over time.

Summary

A GEO audit is a technical and editorial readiness review for AI-driven search. It checks crawler access, renderable and well-structured content, consistent entity information, and visible sources. It’s useful, it overlaps heavily with good SEO practice, and it should never be sold as a prediction of AI visibility.