Why answers need to be quotable
ChatGPT search, Perplexity, Claude and Google’s AI Overviews build their answers from passages they find on the web, and they link to the pages those passages came from. The passages they can use are the ones that answer a question on their own, in the first sentence or two under a heading that asks it. A page can rank well and still never be quoted, because its answers start with an introduction, sit in a paragraph two hundred words long, or only appear once JavaScript has run.
What it checks
It starts with your robots.txt and works out, crawler by crawler, which of the AI companies’ crawlers it lets in. Those split into three kinds: search crawlers, which decide whether an answer can cite you; crawlers that fetch a page when a person asks about it; and training crawlers, which only gather text for training models, so blocking them is a choice that does not keep you out of answers.
Then it reads your pages, from your sitemaps first. Any page carrying nosnippet or max-snippet:0 is flagged, since Google applies those to AI Overviews and AI Mode as well as to its ordinary results, and so is any page with very little text in the HTML the server sends. Finally, every heading phrased as a question is checked against the first thing underneath it: a direct answer of a sentence or two is marked quotable, and anything else gets a specific fix, whether the answer is missing, buried behind an introduction, too long to quote or replaced by a list.
How it differs from tools built on paid data
Tools that pay for search data can match your pages to the questions people actually search for, and list the questions no page of yours answers. This check has no search data, so it works from the questions your own headings already ask. That still covers most of what decides whether a page can be quoted, and the report says plainly which part it cannot see.
Your report and your data
The check is free. It asks for your email address before it runs, and you only join the mailing list if you tick the box. Pages are read by EghosaBot, which obeys robots.txt and its Crawl-delay setting. Each report has its own private link, which works for 90 days and is then deleted along with everything the check gathered. The privacy policy explains the rest.