The Mention Rate is the share of answers in which a brand is mentioned, relative to all answers to a fixed prompt set. If a prompt is asked ten times and the brand appears in four answers, the Mention Rate for this prompt is 40%. Across all prompts of a category or the entire set, this yields the metric for basic visibility in a topic area.
How does measuring the Mention Rate work?
Three steps come first. First: identify typical question patterns. For which prompts do AI systems mention brands at all? They rarely do for general research questions, almost always for recommendation and comparison questions. Second: collect the Mention Rate, several times per prompt and system, because every answer is unique. Third: analyse the Sentiment, because mentioned is not recommended. In addition, the model knowledge is queried: does the system hold all correct information about the brand without web search?
Why does the Mention Rate matter for AI visibility?
Because it answers the question that decides over enquiries: does my company appear when someone asks about my service? The study “AI Blind Test 2026” by CodaAI analysed 7,184 AI answers on 449 mid-sized B2B companies in Germany: in 55% of buyer questions the name of the company under review does not appear, and in 47% the AI recommends competitors instead. Without the Mention Rate, this state remains invisible because it shows up in no website statistic.
What does this mean for your website?
Collect the Mention Rate separately by prompt category and system. A high rate on brand-specific questions and a low one on brand-neutral questions means: the AI knows you but does not recommend you. The lever then lies with third-party sites (Brand Mentions), not with the website. More is not always better: where do I want to be mentioned, and where not? A mention in the wrong context is not visibility.