AI Visibility Audit

Separate retrieval crawlers from training crawlers. Citation requires the retrieval ones to be allowed; training access is a brand policy decision.

A common fatal configuration is blocking the retrieval crawlers that create visibility while allowing the training crawlers — contributing corpus while switching off exposure.

Confirm nothing is blocked at the CDN or edge outside your site config, and that key content does not depend on client-side rendering. What the crawler reads is the HTML you return.

Does the sitemap declare the right domain and every indexable page? On multi-domain multilingual sites, each domain needs to declare its own.

Does structured data appear in server-rendered HTML? If a client-side script injects it, crawlers that do not execute scripts never see it.

Are same-intent pages competing with each other, and how much of the site sits outside the range where ranking produces any traffic?

Layer 3 is mentions and citations: mention rate, average position and the composition of cited sources across a fixed question set.

Layer 4 is conversion: does visibility produce visits, do landing pages answer the intent behind the question, and is the enquiry path intact. Getting the first three right and dropping the fourth is a common and expensive waste.