Free AEO Checker: Score Your Website for AI Answer Engines

What does this AI crawler checker actually test?

It fetches your page exactly the way a retrieval crawler does, with no JavaScript execution, and then grades five things: whether AI crawlers are allowed to reach the page (robots.txt, meta robots, X-Robots-Tag, and CDN-level bot rules), whether your content exists in the raw HTML at all, whether the content can be extracted as a standalone answer, whether your entity and attribution signals are visible, and what structured data is present in the server response. It also shows you the literal text a non-rendering crawler receives, which is usually the most useful part.

Why is the score weighted the way it is?

Because the evidence for each signal is wildly uneven, and most AEO scores pretend otherwise. Crawler access and JavaScript rendering are directly observable facts about whether an engine can read you at all, so together they carry 55 of the 100 points. Answer extractability carries 20. Entity and trust signals carry 15. Structured data carries only 10, and llms.txt carries zero. Every check is labelled proven, contested, or cosmetic so you can see which evidence tier it sits in rather than trusting one composite number.

Do AI crawlers really not execute JavaScript?

Most do not. A searchVIU analysis of 1.3 billion crawler requests found roughly 69% of AI crawlers cannot execute JavaScript at all, and independent testing puts GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Meta-ExternalAgent, Amazonbot and Bytespider in the static-HTML-only group. Only Googlebot, Google-Extended and Applebot render JavaScript. That is why a client-side rendered site can rank fine on Google and be completely invisible to ChatGPT, Claude and Perplexity. It is also why this tool deliberately does not render JavaScript.

Does JSON-LD schema markup help me get cited by AI?

Less than you have probably been told, and the distinction matters. searchVIU ran a controlled test placing facts only inside JSON-LD, and none of ChatGPT, Claude, Perplexity, Gemini or Google AI Mode extracted them on live fetch. Hidden microdata and RDFa were ignored universally too. What engines reliably read is visible HTML text. Schema still has a real job, which is entity disambiguation inside Google's indexing pipeline, where Organization markup with sameAs links helps Google work out which company you are. So implement Organization schema, keep Article author and dateModified accurate, and stop expecting markup alone to earn citations.

Is FAQPage schema worth adding in 2026?

Only when the questions and answers are genuinely visible on the page. Google retired FAQ rich results in May 2026, so there is no search-appearance benefit left, and engines quote the visible answer rather than the JSON-LD wrapper around it. This checker therefore runs a consistency test most tools skip: it compares every question string in your FAQPage schema against your visible text and flags the ones that appear nowhere on the page. Markup-only FAQ blocks are mismatched markup under Google's structured-data policy and earn nothing in AI answers.

Does llms.txt do anything?

There is no evidence that it does, which is why this tool reports it and scores it at zero. Google has said explicitly that Search does not use llms.txt, no major answer engine documents it as a ranking or citation input, and published analyses of sites that added one have not isolated a measurable effect. It is cheap agent-readiness housekeeping and fine to publish, but it is not a substitute for being crawlable, server-rendered and worth quoting. Any checker that gives llms.txt heavy weight is scoring the easiest thing rather than the thing that matters.

What is the difference between GEO, AEO and AI SEO?

In practice they are the same discipline under three names. GEO (generative engine optimization) is the most common term in academic and agency usage, AEO (answer engine optimization) is more common among marketers, and AI SEO tends to be used by people extending existing SEO practice. All three describe making your content findable, readable and quotable by systems that answer questions directly instead of returning a list of links. This tool covers the technical half of all three.

Which AI crawlers should I allow, and which can I block?

Separate retrieval from training, because they have different consequences. Retrieval and citation bots (OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User, PerplexityBot, Perplexity-User, Googlebot, Bingbot, DuckAssistBot) fetch pages to build answers, so blocking them removes you from AI results. Training bots (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended, cohere-ai, Bytespider) feed model training, and blocking them is a legitimate commercial choice with no visibility cost. This checker reports them separately and only penalises blocked retrieval bots.

My robots.txt allows everything, so why would I still be blocked?

Because robots.txt is not the only gate. CDN and WAF bot-management rules sit in front of your origin and reject requests before robots.txt is ever read. Cloudflare's 'Block AI bots' toggle, AWS WAF Bot Control and similar features are commonly enabled by someone on the infrastructure side without the marketing team knowing. This tool probes your page twice, once with an ordinary user-agent and once carrying a bot token, and flags any divergence in the HTTP status. That is the block most audits miss.

Does a good score mean AI will cite me?

No, and it is worth being blunt about that. This measures eligibility, not citation. Everything here removes technical reasons an engine cannot read or quote you, which is necessary but nowhere near sufficient. Whether you actually get named in answers depends mostly on content quality, topical authority and third-party corroboration such as being mentioned on Reddit, in review roundups and by other sites. Readiness and citation are separate measurements, and the gap between them is not closed with more technical work.

How often should I re-run this?

Re-run it after any deploy that touches rendering, routing, robots.txt or your CDN security settings, since those are where silent regressions happen. Beyond that, monthly is plenty. The fixes themselves take hours, but engines do not re-crawl on demand and several build answers from index snapshots refreshed on their own schedule, so expect crawler activity within days and any change in citation over four to eight weeks.

Is my data stored?

No. The checker fetches the URL you enter, analyses the response in memory, returns the result, and keeps nothing. There is no account, no email gate and no stored history, so re-running a check always performs a fresh fetch.