GEO audit

A GEO audit that checks readiness, not promises

Generative engine optimisation (GEO) readiness checks run with every SEO Sentinel audit. They look at what you can observe on the site itself: which AI crawlers robots.txt allows, how content is structured, whether the brand is described consistently, and whether articles show authors, dates and sources.

Explore Interactive Demo Request Beta Access

Readiness is not visibility

A GEO audit can tell you whether AI search crawlers are allowed in and whether pages are easy to parse and attribute. It cannot tell you whether ChatGPT, Gemini or Perplexity will mention or cite the site. SEO Sentinel says so on every GEO report, and there is no GEO score.

Google states that its AI features in Search have no additional requirements beyond normal Search eligibility, so most suggestions here are simply good practice for readers and search in general. That is how we present them.

The checks reuse the same crawl as the technical audit: no extra page requests, no AI API calls and nothing sent to any AI service.

GEO / AI Readiness report for a fictional roofing website: pages analysed, search crawlers allowed, entity and content findings, and a limitations list stating that the checks cannot predict AI citations. — Screenshot from the public interactive demo. The agency and websites are fictional.

AI crawler access, labelled by purpose

robots.txt rules are tested for 13 documented crawlers and tokens on up to 20 important URLs, using standard RFC 9309 precedence: a crawler’s own group replaces *, the longest matching rule wins, and the rule’s line number is shown as evidence.

Crawler purposes checked by the GEO audit
Purpose	Examples	How SEO Sentinel treats it
AI search	OAI-SearchBot, Claude-SearchBot, PerplexityBot	Blocking these is reported, because it can keep a site out of those AI search features.
Search engine	Googlebot, Applebot	Googlebot also crawls for Google’s AI features in Search.
AI training	GPTBot, ClaudeBot	Blocking training is a business decision, not an error.
Training opt-out token	Google-Extended, Applebot-Extended	robots.txt tokens only; no separate crawler visits.
Dataset	CCBot (Common Crawl)	Builds an openly available web dataset.
User-triggered fetcher	ChatGPT-User, Claude-User, Perplexity-User	Their operators say robots.txt may not apply to user-initiated fetches.

The crawler registry is checked against each operator’s own documentation. Vendors add and rename crawlers, so it is maintained over time. Read the AI crawler robots.txt guide.

The other three readiness areas

Content readiness

Non-descriptive H1s and subheadings
Long text without subheadings; very long paragraphs
Question headings with no answer below them
Pages that look like JavaScript shells
Snippet restrictions (nosnippet, max-snippet:0, data-nosnippet), which also limit Google’s AI features

Pages are never flagged just for having no FAQ or question headings.

Entity signals

Conflicting organisation names across structured data, og:site_name, titles and the copyright line
No Organization markup; no About or Contact link
A structured-data phone number that differs from the page’s tel: link
Structured contact details not visible on the page

Sources & evidence

Article pages (Article, BlogPosting or NewsArticle markup) missing an author or date
Inconsistent published/modified dates
Articles with no external sources
Quotations without visible attribution

Every GEO finding has a severity (never a “confirmed error”), a verification level (confirmed, likely or unverified) and a scope: sitewide/template when it repeats on at least 5 pages and 60% of comparable pages, otherwise page-specific.

What it does not do

robots.txt permission doesn’t prove access. CDNs, firewalls and bot protection can still block AI crawlers. Refused requests during the crawl are counted as a hint only.
No measurement of AI mentions or citations. For recorded observations, see AI visibility tracking.
No JavaScript rendering, and heading/answer heuristics are English-oriented.
Entity and About/Contact findings depend on crawl coverage, and a brand shown only in a logo image can’t be compared.

Learn

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.

Draft · under review

AI crawlers and robots.txt: what each token controls

Which robots.txt tokens control AI search, AI training and user-triggered fetchers (GPTBot, OAI-SearchBot, ClaudeBot, Google-Extended) and how to test them.