How Peekaboo measures AI visibility.

The basic observation is a prompt run on a named AI surface. A run records the prompt, model or surface, collection time, returned answer, detected brand appearances, position or score where available, and cited sources.

Prompt wording, market, language, account state, model updates, and answer randomness can change the output. Comparisons are most useful when those conditions and the prompt cohort remain consistent.

Peekaboo aggregates eligible run scores on a 0–100 scale for the selected brand, time range, prompts, models, and filters. Every eligible run remains in the denominator. When the brand is not mentioned, that run contributes zero rather than being excluded from the average.

The displayed run count and filters are part of the result. Scores from different prompt sets or collection contexts should not be treated as directly equivalent without reviewing those differences.

• A mention means the answer refers to the tracked brand.
• A recommendation requires language that presents the brand as an option.
• An owned citation links to a domain controlled by the tracked brand.
• A third-party citation can discuss a brand without citing the brand’s domain.

A visibility change does not by itself prove a content change caused the result, that a user saw the answer, or that the answer generated revenue. Teams should evaluate repeated observations, search and referral data, and qualified business outcomes together.

AI providers can change models, interfaces, citations, and access behavior without notice. Peekaboo reports the observable context available at collection time and corrects documentation when the implementation changes.

Report a methodology question or suspected error to team@aipeekaboo.com. Editorial corrections follow our published policy.