Last updated:
LLM SEO Checker: AI Readiness Report
This LLM SEO checker tests whether AI search crawlers can reach a page and what they find there.
Our server fetches the page, the site's robots.txt and its llms.txt as SchemaBizBot; everything is checked here in your browser.
How to use the LLM SEO checker
Enter a page's address. Our server fetches the page, the site's robots.txt and its llms.txt as SchemaBizBot, and the report is built in your browser: a score out of 100, up to five fixes in order of what they're worth, every check with what it found, and a table of which crawlers may fetch the page. A link such as /ai/readiness-checker/?url=https://example.com/page runs the check when it opens.
"Try an example page" loads a beginner's guide on a trail running blog, without fetching anything. Its robots.txt blocks PerplexityBot and Claude-SearchBot from /guides/, so two of the 8 AI search crawlers can't index the page, and that's the top fix. Its BlogPosting names an author but no dates, and its publisher has no logo or sameAs links. GPTBot is blocked too, but that's a training crawler: the report notes it and the score ignores it. The site has no llms.txt, which is a note.
What the report checks, and why
Google's guidance on AI features is short: to appear as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown with a snippet, and there are no other technical requirements. It also says you don't need AI text files or special structured data. Other AI search engines index the web with their own crawlers, which robots.txt controls by name. So the score weighs access and indexing most:
- The page loads (15 points): a 200 answer.
- Indexable by Google (20): no
noindexfor Google in a robots meta tag or theX-Robots-Tagheader. A rule for another crawler, such asX-Robots-Tag: otherbot: noindex, doesn't count against it. - Snippets allowed (10): no
nosnippetormax-snippet:0; a very smallmax-snippetcosts half. Parts markeddata-nosnippetare noted: they stay out of snippets and AI features. - Googlebot and Bingbot may crawl it (15): the page's path tested against robots.txt, the way the robots.txt checker does.
- AI search crawlers may crawl it (15, shared among them): OAI-SearchBot, Claude-SearchBot, PerplexityBot, meta-webindexer, Amzn-SearchBot, MistralAI-Index, DuckAssistBot, YouBot, each checked on its vendor's own documentation.
- Valid structured data (10): JSON-LD, Microdata or RDFa without errors, checked by the same engine as the Schema Markup Validator. Markup isn't required for AI features, but it says what the page is without guesswork.
- Who's behind it (5): an Organization or Person with a name, a logo or image, and
sameAslinks to its official profiles. - Author and dates (5, articles only): an
author,datePublishedanddateModified.
The score is the points earned out of the points that apply, so a product page isn't marked down for having no author. A page that doesn't load isn't scored at all: most checks need the page. Two things are reported and never scored: training crawlers, because blocking them is a decision about model training rather than search, and llms.txt, because it's an unconfirmed proposal. When robots.txt or llms.txt couldn't be fetched just then, the report says so and leaves them out.
Fixing what the report finds
- Blocked AI search crawler. Add an
Allowfor its token, or remove theDisallowthat catches it; test the result in the robots.txt checker. - noindex or nosnippet. Usually a template or plugin setting left on from staging. The meta tag checker shows every robots tag and header on the page.
- No structured data, or errors. Generate markup for the page's main entity with the JSON-LD Generator, then validate it.
- No publisher identity. One Organization node with its
logoandsameAslinks, referenced from every page, is enough; the Organization schema generator writes it. - No author or dates. Add them to the Article or BlogPosting. The AI Search Optimizer scores the markup's entity and citation signals in more detail.
What this checker can't tell you
Whether a page is cited depends on the question, on what other pages say and on choices AI companies don't publish. The checker can't see those. It also reads the page as SchemaBizBot sees it: a site that treats unknown crawlers differently, or a firewall that challenges bots, may answer AI crawlers differently from us. Treat the report as a checklist of the parts you control.
Frequently asked questions
FAQ
Does a high score mean AI assistants will cite my page?
No. No AI company publishes how it picks the sources it cites, and nobody can promise a citation. The score measures what is documented and in your control: whether crawlers may fetch the page, whether Google may index it and show text from it, and whether its structured data is valid and says who's behind it. A low score points at something that can keep a page out; a high one only means nothing documented is in the way.
Why doesn't blocking GPTBot or ClaudeBot lower the score?
They collect content that may be used to train models; blocking them is a choice about training, not about search. AI search uses separate crawlers, such as OAI-SearchBot, Claude-SearchBot and PerplexityBot, and those count. Google-Extended works the same way: Google says it doesn't affect Google Search, and its AI features use what Googlebot may crawl.
Why doesn't llms.txt count toward the score?
It's a proposal, and support from the major AI providers is limited or unconfirmed. Google says no AI text files are needed for its AI features. The report says whether the site has one and whether it follows the format, and links to the llms.txt generator if you want to add one.
Does it check pages that add their markup with JavaScript?
It reads the HTML the server sends, as the markup checkers do, without running scripts. If a page adds its structured data with JavaScript, the report won't see it; paste the rendered HTML into the Schema Markup Validator to check that markup.
Does the checker store the page?
No. Our server fetches the page, the site's robots.txt and its llms.txt as SchemaBizBot and sends them to your browser, where the report is built. Like every tool on the site, the checker records what kind of check ran: the score, how many crawlers are blocked, whether the page is noindex or has structured data, and which checks need work, never the page or its markup. The page's address is recorded without its query string. See the privacy policy.