Site readiness — can an engine read you at all?
Before an engine can recommend you it has to reach your site, read it without running your JavaScript, and find a sentence that says what you are. Site readiness runs fifteen checks the way a crawler does — raw HTML, no JavaScript, no login — and scores them out of 100 across four groups: access, content, markup, and the off-site half. Every failing check shows what we saw and the fix, ready to paste.
How it works
Access: can they get in?
Homepage answers a plain request; robots.txt allows the eight AI crawlers we check for (GPTBot, ClaudeBot, PerplexityBot, Google-Extended and the rest); the content is present without JavaScript; a sitemap is published.
Content: does it say what you are?
Title and description say what the product is; one clear H1; a plain statement of what it is, for whom, at what price, against which alternatives — judged by a model reading your page the way an engine does; comparison pages; a visible price.
Markup and the off-site half
Organization, Product or SoftwareApplication, and FAQ schema; llms.txt; canonical and Open Graph tags. Then, from your scan: whether your own pages get cited at all, and which third-party pages the engines trust instead.
In depth
The fix is written, not described
A failing schema check hands you the JSON-LD with your name, category and price filled in. A robots block hands you the lines to add. A missing llms.txt hands you the file. Copy, paste, re-check.
Weighted by what actually matters
Robots access and the plain statement carry the most weight because they decide whether anything else is read. llms.txt carries almost none, and the check says why: no public evidence yet that the big engines read it.
Feeds the action plan
Failing checks become tasks at the top of the action plan, above the scan's moves, because a site the engine can't read makes every other move moot. The check re-runs when the audit is a day old, so a fix you ship drops off the plan by itself.
Free without an account
The same fifteen checks, minus the model call, run on any domain at citegraph.app/readiness: score, blockers, a share link and an image for the result. The pasteable fixes are the part behind the login.
What we won't claim: that a 100 gets you recommended. Readiness is the half you control; it makes you readable, not chosen. The scan measures the other half — whether the engines actually name you — and the off-site checks say so in the score.
Common questions
What does the site readiness check look at?+
Fifteen checks in four groups: access (homepage reachable, AI crawlers allowed in robots.txt, content present without JavaScript, sitemap), content (title and description, one H1, a plain statement of what, for whom, price and alternatives, comparison pages, visible pricing), markup (Organization, Product and FAQ schema, llms.txt, canonical and Open Graph) and off-site (whether your own pages get cited, and which third-party pages the engines trust).
How is it different from an SEO audit?+
It checks the things an answer engine needs rather than a ranking algorithm: whether the AI crawlers specifically are allowed, whether the page makes sense with no JavaScript, and whether there is a sentence an engine can quote that says what you are. Page speed, backlinks and keyword density are not in it.
Is the free public checker the same?+
Same engine, same fifteen checks, two differences: the public version skips the model call that judges the plain statement (it uses a pattern check and says so), and it shows the fixes' names but not the pasteable code.
See it on your own domain — first dossier in about a minute.