Loading…
Loading…
Tools field guide
Run a free AEO audit (GEO audit): check AI crawler access, structured data, answer-ready content and the outside sources AI trusts, then fix what matters.
By AnswerMentionsReviewed Oct 17, 20268 minute read
Written for: Site owners, SEOs and agencies who want to know what stops AI engines from citing a site, with the evidence behind each check.
Short answer
An AEO audit, also called a GEO audit, checks whether AI answer engines can reach your site, understand what you sell and quote you, and whether the outside sources they trust describe you well. Enter your site above: the free audit runs eight site checks and asks ChatGPT three buyer questions in about 90 seconds.
Our position
Most AEO audits stop at your own site: schema, llms.txt, headings. That is the easy half. In our study, 93% of the pages AI cited for buying questions were lists, so an audit that never looks off-site grades the wrong exam.

An AEO audit tests whether answer engines such as ChatGPT, Perplexity and Google’s AI Overviews can use your site as a source and recommend you. GEO, generative engine optimization, is the same idea under its research name. A useful audit covers four layers: access, understanding, answerability and outside trust.
GEO comes from a 2023 paper by Aggarwal and colleagues (KDD 2024). If an agency sells you a “GEO audit”, ask for the same four layers.
An AEO audit runs fourteen checks across four layers. Access covers crawler rules, firewalls, Google’s AI opt-out, indexing and JavaScript. Understanding covers entity markup, product facts and consistency. Answerability covers answer-first sections, question-style content and freshness. Outside trust covers the third-party pages AI cites for your category.
| Layer | Check | Why it matters | In our audit |
|---|---|---|---|
| Access | AI search crawlers allowed in robots.txt | A blocked crawler can’t fetch your pages for answers | Free (robots.txt); $29 report checks 10 AI crawlers |
| Access | Firewall or CDN not blocking bots | Bot protection can block crawlers robots.txt allows | $29 report |
| Access | Google’s AI control set to include | Excluded sites get no traffic from AI Overviews or AI Mode | Manual: Search Console |
| Access | Indexable, not noindex | Google’s AI features only use indexed, snippet-eligible pages | Free |
| Access | Key content in the HTML | Many AI crawlers don’t run JavaScript | $29 report |
| Understanding | Organization markup | Tells engines who you are and which profiles are yours | Free |
| Understanding | Product or service markup | Machine-readable facts about what you sell | Free |
| Understanding | Same facts on every profile | Conflicting facts make you a riskier answer | $29 report |
| Understanding | llms.txt (optional) | A cheap map of your key pages; Google says Search ignores it | Free |
| Answerability | Answer-first sections | Engines lift self-contained answers | $29 report |
| Answerability | Question-style content | Matches how buyers phrase questions | Free |
| Answerability | Depth and freshness | Thin or stale pages give nothing to quote | Free (homepage) |
| Outside trust | Listed where AI looks | Lists and reviews decide most shortlists | Free preview; full in $29 report |
| Outside trust | Earned media and reviews | Independent descriptions carry the most weight | $29 report |
Check robots.txt for each AI crawler by name, then check your CDN or firewall, because bot protection can block a crawler your robots.txt allows. Search crawlers decide whether you can appear in answers; training crawlers decide whether your content trains future models. Treat them as separate decisions.
OpenAI documents its crawlers as separate settings and says robots.txt changes take about 24 hours to apply. Google says Google-Extended doesn’t affect inclusion in Google Search, and its AI features draw on the normal Search index.
Google added a third switch in 2026: the Search generative AI control in Search Console, live for all sites since August 31, 2026. It defaults to include. Sites that exclude themselves get no traffic or impressions from AI Overviews or AI Mode, so check nobody flipped it.
On Cloudflare, check the AI bot policy: since September 15, 2026, new domains block Training and Agent bots on pages with ads by default; Search bots stay allowed.
Then check what a crawler receives. In Vercel’s analysis of AI crawler traffic, OpenAI’s, Anthropic’s and Perplexity’s crawlers didn’t render JavaScript; Google’s and Apple’s did. Prices or features that only appear after scripts run may be invisible to the first group.
| Crawler | Company | What it does | If you block it |
|---|---|---|---|
| OAI-SearchBot | OpenAI | Fetches pages for ChatGPT search | You drop out of ChatGPT search answers |
| GPTBot | OpenAI | Collects training data | No effect on ChatGPT search, per OpenAI |
| Claude-SearchBot | Anthropic | Indexes pages for Claude’s search | Claude’s answers may not show your pages |
| PerplexityBot | Perplexity | Indexes pages for Perplexity answers | You stop appearing as a Perplexity source |
| Googlebot | Search indexing, including AI Overviews and AI Mode | You leave Search and its AI features | |
| Google-Extended | Controls Gemini training and grounding in the Gemini app | You stay in Google Search; Gemini app answers may lose you | |
| Applebot | Apple | Search in Spotlight, Siri and Safari | You leave Siri and Spotlight answers |
| Bingbot | Microsoft | Bing index used by Copilot | You leave Bing and Copilot answers |

Give engines one unambiguous description of your company and products: Organization markup with your official name, logo and profiles, Product or Service markup with real prices, and the same facts on every profile you control. Markup has to match what the page shows. It helps engines read you; it doesn’t make them pick you.
Google says there is no special markup required for its AI features. In Ahrefs’ May 2026 test of 1,885 pages that added JSON-LD against 4,000 controls, schema had no clear effect on ChatGPT or AI Mode citations. Organization markup still helps engines disambiguate your organization.
llms.txt is a 2024 proposal for a plain-text map of your key pages. Google’s AI optimization guide says Search ignores such files, and in Ahrefs’ server logs from 137,000 domains, 97% of llms.txt files got no requests in a month. Score it as optional.
Write pages so any section can be quoted on its own: a direct answer in the first sentence or two, then the detail. Use the words buyers use, put numbers and dates in sentences, and keep facts current. In the GEO study, adding quotations, statistics and cited sources lifted visibility by roughly 28–41%.
Keyword stuffing scored below baseline in the same study. Freshness counts too: across 17 million citations, Ahrefs found AI assistants cited content 25.7% fresher than organic results, with ChatGPT the most freshness-hungry.

AI engines lean on what others say about you more than on your own pages. Earned media made up 84% of AI citations in Muck Rack’s study of 25 million links, and in our study of software questions, 93% of cited pages were lists. An audit that never checks those sources misses where most recommendations are decided.
Muck Rack’s data covered ChatGPT, Claude and Gemini; our study found cited lists spread over 193 domains, so the sites that matter depend on your category.
The audit’s Missing Source Map lists the sites AI cited when it named a rival and left you out: your outside-trust fixes.
An SEO audit asks whether Google can crawl, index and rank your pages. An AEO audit asks whether AI engines can use and recommend you, which adds AI crawler rules, answer-ready writing and off-site sources. The overlap is large, because Google’s AI features only use pages that are indexed and eligible for snippets.
Google’s own guide calls optimizing for its generative AI search “still SEO”. Start with the AEO audit if buyers ask AI before they search; our AI visibility audit measures the outcome side.
| SEO audit | AEO audit | |
|---|---|---|
| Question | Can Google rank my pages? | Will AI engines cite and recommend me? |
| Unit of success | A page’s position | Your brand inside an answer |
| Unique checks | Core Web Vitals, internal links, rankings | AI crawler rules, answer-first writing, cited sources |
Enter your website in the form at the top of this page. The audit fetches your homepage, robots.txt, sitemap and llms.txt, runs eight readiness checks, then asks ChatGPT three buyer questions and shows who it names and which sources it cites. The $29 report adds four more engines and a five-layer site audit.
Your domain or brand in the form above. No signup or card.
Eight homepage checks, then three live ChatGPT answers with brands and sources.
Access first, then outside sources, then markup and content.
Test your own brand
Ask ChatGPT three real buyer questions about your market and see who it recommends, free, in about 90 seconds.
Questions that change the decision
No. llms.txt is a proposal, and Google says its Search ignores it. It is cheap to add, so add it if you like, but crawler access, clear pages and outside mentions matter far more.
That is a business decision about training, not search. OpenAI controls ChatGPT search with OAI-SearchBot, a separate crawler, so you can block GPTBot and still appear in ChatGPT search answers as long as OAI-SearchBot is allowed.
It helps engines read your facts correctly; it doesn’t make them recommend you. Use Organization and Product markup that matches the page, then spend most of your effort on the third-party lists and reviews AI actually cites.
Every quarter, and after any migration or CDN change. Access problems appear overnight; content fixes take weeks to show.
Nothing practical. GEO is the research term from the 2023 paper; AEO is the marketing term for the same work.
Platform documentation supports factual statements. Where we describe an audit method or prioritization rule, that is AnswerMentions' operating judgment and is labeled as such.
Continue the diagnosis