Measure AI visibility with evidence you can explain

Readiness describes the site’s technical access, content structure, query coverage, and supporting authority signals. It helps identify work the business can control, such as an inaccessible page or an answer that lacks clear evidence.

A high score does not establish that Google has indexed a page or that an AI system has cited it. Re-running an audit after a fix can show that the audited condition changed. Search and citation outcomes need their own evidence.

The right source depends on the question. Avoid combining different kinds of evidence into a single claim of success.

• Search Console: impressions, clicks, queries, and pages in a stated search window.
• Google Analytics: recorded traffic and configured events, subject to the quality and availability of tracking.
• Sampled AI checks: what a named engine returned for a particular question at a particular time.
• Imported AI observations: citations or mentions reported by the connected source, with its own coverage and dates.
• Local evidence: Google Business Profile reporting where the business has connected it.

Ask which engine, which question, which date, and which source produced the observation. A mention is not always a linked citation. A sample of answers is not a census of every answer a potential customer could see.

If a report gives a rate, keep the number of checks and the sampling conditions beside it. A new engine or a different question set changes the comparison. An older imported citation and a fresh model check should not be blended into one current visibility claim.

Treat the result as unavailable or limited. A disconnected account, a failed request, and a measured zero are different situations. Before deciding that a page failed, check whether the source covers that page and whether the observation is recent enough to answer the question.

Connected reporting requires authorized access to the relevant property and a plan that includes it. The free website check does not connect your private analytics accounts. A report can still be useful when it states what is known and what needs to be measured next.

It shows what changed across the selected windows. By itself, it does not prove that one edit caused the result. Other releases, demand changes, indexing delays, tracking changes, and engine variability can affect the comparison.

Keep the conclusion proportional to the evidence. Say the technical issue was resolved when you have verified the fix. Say clicks or citations increased only when comparable observations support it. No ranking, recommendation, citation, or growth outcome is guaranteed.

The public sample report separates readiness, search evidence, model checks, and imported citations. It is a fixed August 2026 snapshot of an Awesome.Digital-owned property. Use it to inspect the reporting approach, then run a new check for evidence about your own site.