Mention rate is one of the simplest metrics for putting a number on a brand’s AI visibility. It addresses a straightforward question: In how many answers does a brand come up when people ask AI systems about its field? For example, ten questions, each put to an AI system ten times, produce 100 answers. If 30 of them name the brand, the mention rate is 30%.
How mention rate is measured
Measurement starts with a prompt set, a fixed set of typical questions, known as prompts, that customers might ask an AI system. For each answer, the only thing recorded is whether the brand appears, not how often. To catch every mention, the analysis needs a list of all the brand’s names, for example the company name with and without its legal form, plus key product names.
Because AI answers vary from run to run even for the same question (answer variability), a single query says little. In prompt tracking, each question is therefore asked repeatedly, and the results are combined over a longer period. What matters most is the trend over months, reported separately for each AI system, since ChatGPT, Gemini, Claude, and Perplexity answer differently.
What mention rate shows and what it doesn’t
The type of question matters. When a prompt already names the brand (for example, “What does Company X offer?”), the answer usually names it too. Only questions about a category or a need without a brand name show whether an AI system brings up the brand on its own. The two types of questions are therefore analyzed separately. A September 2026 preprint on product recommendations from six language models without web access found that even some large, well-established brands were never recommended when the question named only a product category.
Mention rate only counts whether a brand appears. It does not show where in the answer, in what tone, or with which facts the brand is named. Nor does it capture whether the answer gives one of the brand’s pages as a source: such AI citations are counted separately, for example as citation rate. Mention rate is also the basis for comparisons: the mentions of several brands for the same questions yield their AI share of voice—each brand’s share of all brand mentions in its competitive field.
For GEO work, mention rate is a good entry point to GEO monitoring because it is easy to explain, including to management. It stays comparable over months as long as the measurement setup is unchanged: the same questions, AI systems, number of runs, and name variants.