Useful because it says where the read stops.

Start with the limit

What it cannot see

Private evidence

It cannot access analytics, CRM records, experiments, sales calls, or authenticated product flows.

Actual causality

It can identify friction and weak signals. It cannot prove a page change caused a conversion result.

The whole market

Engine answers are sampled prompts at one moment. They are not a complete ranking or every buyer’s opinion.

Hidden or changing pages

It reads what can be reached publicly during the run. Personalization, paywalls, and later changes sit outside that evidence.

The read

How it works

The reader captures the public page, gathers reachable evidence, asks a bounded set of buyer questions, then compiles only the evidence that landed.

A finding is a directional decision aid. Revenue language uses stated assumptions; it is never reported as observed performance.

Three lenses, one read

What each lane measures

Your website

Public page clarity, visible decision friction, page structure, and performance evidence that could affect a first visit.

AI visibility

Whether sampled AI answers cite or recognize the business, plus reachable technical and content signals behind that result.

Brand

How the offer is differentiated in the observed evidence, including likely category confusion or competitor collision.

Incomplete means incomplete

What the completion notice reports

Every read ends with a completion record. If a leg did not land, the report names the missing category and its returned reason.

• Page captures, including desktop or mobile capture attempts.
• Evidence collection, including named public-site views and files.
• PageSpeed and Core Web Vitals retrieval.
• AI-engine question runs.
• Read or compilation work when a later synthesis step cannot finish.

Missing evidence is not turned into a zero, a passing score, or a made-up explanation.