How to interpret the AI Visibility Score

When an answer mentions one or more distinct entities, Peekaboo assigns each detected entity a position-based score:

totalMentions is the number of distinct detected entities after entity resolution, and validRank is the entity’s position constrained to that range. Rank 1 is the first detected entity. The formula gives the top position 100 and reduces the score for later positions. The result is clamped to the 0–100 range. An entity that is not mentioned receives zero for that run.

An answer contains five distinct brand mentions. Your brand appears third:

Position is only one dimension of the answer. Read the answer, recommendation language, sentiment, and cited sources before deciding what the position means.

For a selected window, Peekaboo takes the best applicable score for the tracked entity in each eligible run, uses zero when it was not mentioned, then averages all of those run scores. The same aggregation logic applies to the tracked brand and its competitors.

A dashboard should present the second state as unavailable—not silently turn it into a measured zero. Always report the run count with the score.

A score is only comparable when the measurement frame is comparable. Check:

• the prompt cohort and whether prompts were added, removed, or rewritten;
• the AI models or surfaces included;
• language, location, and other collection context;
• the start and end dates and number of eligible runs; and
• topic, intent, model, or other active filters.

A score change can come from a real answer change, normal model variation, a model update, or a changed measurement frame. Break the aggregate down before acting.

The score summarizes observed prominence across a defined test. It is useful for tracking direction, comparing entities under the same conditions, and finding prompts or models that deserve investigation.

It is not market share, audience reach, a traffic forecast, a probability that a person will see the brand, or proof of revenue. There is no universal “good” score across different industries and prompt portfolios; your consistent baseline and competitive context are more informative than a generic benchmark band.

For the collection context behind the metric, read How Peekaboo measures AI visibility.