What we check.

Whether robots.txt lets in the crawlers that fetch pages for AI answers — ChatGPT search, Claude, Perplexity, Google, Bing and others. Each one blocked is an assistant that cannot use the page at all.

Whether training crawlers such as GPTBot, ClaudeBot and Google-Extended can read the page. Blocking them is a legitimate choice, so it only ever lowers the score a little.

Whether a noindex or nosnippet directive (in the page or the X-Robots-Tag header) keeps the page out of search results and the AI answers built on them.

How much text is in the HTML your server sends. Many AI crawlers do not run JavaScript, so a page built in the browser can look empty to them.

If your server or a firewall answers our crawler with 401, 403, 429 or 503, AI crawlers that identify themselves are likely to get the same answer.

A single main heading that says what the page is about.

H2 and H3 subheadings that break the page into parts an assistant can quote on their own.

Headings phrased as the questions people type into AI assistants, with the answer directly underneath.

Enough specific, factual text in the main content to answer a question without the reader going elsewhere.

Steps, prices, features and comparisons in a form that is easy to extract.

Whether the page describes itself in JSON-LD, the structured data format search engines recommend.

Whether every JSON-LD block is valid JSON. A trailing comma is enough for crawlers to ignore a block.

Organization, LocalBusiness or Person schema, ideally with sameAs links to your official profiles so assistants can connect your brand across the web.

Schema for the page itself: Article, Product, Service, FAQPage, HowTo, Event and similar types.

Present, and 20–65 characters.

Present, and 50–170 characters. Search results often show it, and AI search tools pick their sources from those results.

A rel="canonical" link, so duplicate addresses are credited to one page.

og:title, og:description and og:image, for a proper preview wherever the link is shared.

A lang attribute on the <html> element.

A byline, an author meta tag or author/publisher in the schema — who stands behind the content.

A published or updated date, and whether it is within the last year.

Links from the main content to the external sources behind your claims.

Whether the page is served securely.

A sitemap at /sitemap.xml or listed in robots.txt, so crawlers can find all your pages.

A Markdown guide to your site at /llms.txt. It is an emerging convention and support from the major assistants is not confirmed, so it carries little weight.

How quickly the HTML arrives at our crawler. Slow pages risk timing out for crawlers with tight limits.