Insights · Field note · Growth and AI visibility
Reading an AI visibility report: a guide for executives
What each figure in the monthly evidence report means, what to be suspicious of, and the one question to ask about every number.
An AI visibility report should answer four things in order: visibility (share of voice and citations by engine against the baseline and competitors), demand (qualified inquiries by source), economics (cost per qualified inquiry and the share of demand arriving without paid spend) and asset (pages published and refreshed, authorship coverage, site performance). Every figure should carry a source and a method. The one question to ask of each number is: how was this measured, and would it look the same if someone else measured it. A report that cannot answer is a brochure.
In detail
What the evidence shows.
Visibility
Share of voice is a frequency across repeated runs of a defined prompt set, by engine. Ask how many runs, how many prompts, whether the set changed, and how competitors were measured. A single-engine figure or a set chosen after the fact is a warning sign. The trend against the baseline is the point; the absolute number is less meaningful than its movement.
Demand
Qualified inquiries by source, with the definition stated and the counts reconciled against the inquiries the business received. Be suspicious of platform conversions, sessions and traffic presented as demand. Referred visits from answer engines understate their effect, because the cited organization is often contacted directly; the report should say so.
Economics
Cost per qualified inquiry by source, paid spend trend, and the earned-demand share: what proportion of inquiries would continue if spend stopped. This is the figure that decides budgets. Ask whether the paid manager’s fee depends on spend; if it does, read the recommendation accordingly.
Asset and the evidence index
Pages published and refreshed, authorship coverage, citations earned by page, performance and security posture of the site. Then the evidence index: the source for every figure. Read a bad month with respect; a report that states bad news with reasons is one you can trust in a good month. See monthly evidence report and monthly evidence report versus marketing dashboard.
Sources
Primary references.
- Performance Data engagement observations (first-hand: David Selakovic); method described on the Measurement and evidence pages.
- Semrush, 2025. AI Mode and AI Overviews study: about 69 million U.S. desktop sessions, May to July 2025; AI-search visitor conversion analysis.
By David Selakovic · Published 2026-10-16 · Revised 2026-10-16 · Reviewed every 180 days