Brand mention probability is the percentage of repeated AI-search responses to a defined prompt set in which a specified brand is named, whether or not it is recommended or cited. It is a metric defined by blimpp, published here with its full measurement method. It measures the floor of visibility: whether the engine reaches for your name at all when answering in your category.
In one sentence
Mention probability tells you how often you exist in the answer, before any question of whether the answer is kind to you.
How brand mention probability is calculated
Brand mention probability = (responses naming the brand ÷ total responses tested) × 100
A response counts once however many times it names the brand, and namings count regardless of sentiment or advisory force; what qualifies as a mention versus a recommendation follows the measurement methodology. Sampling follows the repeated-run protocol. It differs from AI share of voice in its denominator: probability is measured against responses, share of voice against all brand namings.
Worked example
A 20-prompt category set is run five times each on one engine, producing 100 responses. The brand is named in 41 of them: mention probability 41%. In the same responses it is recommended in 12, so a 29-point gap separates being known from being advised. Illustrative example: blimpp benchmark data is added to this page as studies publish.
Why brand mention probability matters
Engines cannot recommend what they do not name, so mention probability is the precondition every other outcome sits on. It is also the sharpest diagnostic when read against recommendation rate. Near-zero mentions is an association problem: the engine does not connect the brand to the category, and the work is footprint and entity building. High mentions with low recommendations is an evidence problem: the engine knows you and does not advise you, and the work is what independent sources say. The two problems are routinely confused and their fixes do not overlap.
What affects brand mention probability
Evidenced factors: the brand’s presence across the sources engines retrieve for the prompt class, since names enter answers through retrieved material and trained association. Factors with practitioner evidence: strength of entity association between the brand and the category, distinctiveness of the brand name, and breadth of third-party coverage. These are treated as working hypotheses until tested.
Related concepts
References
- Martinez, O. (2026), Optimizing Visibility in Generative Engines: A Critical Survey of GEO: arxiv.org
Author: Harpal Singh · Last reviewed: 7 August 2026