Quick answer
BrandWater defines visibility as mentions of a brand divided by eligible observations, the same prompt set run repeatedly across ChatGPT, Gemini, Perplexity and Claude, with the sample size always shown. Two short videos cover the method itself and what a first free audit looks like.
Our methodology page spells out every definition, sampling rule and denominator in writing. These two videos cover the same ground out loud, for anyone who would rather watch it once before reading the detail.
The full methodology, in three and a half minutes
The core formula is simple on purpose: mentions of the brand divided by eligible observations, where an eligible observation is one run of one prompt, on one platform, that could plausibly have named the brand. We run a defined prompt set repeatedly, not once, because an individual AI answer is sampled, not looked up, and a single run can be typical or an outlier.
What a first audit actually shows
A free AI visibility check runs this same method once, on a smaller scale, so a brand can see its first real mention rate, citation rate and share of voice before deciding whether to track it properly.
Read the written version
Cite this page
BrandWater AI Research. (2026, 18 October 2026). How we measure AI visibility: the methodology, on video. https://brandwaterai.in/blog/how-we-measure-ai-visibility-on-video
How we work
Figures are dated and linked to their sources. Where none exist we say so. Read our methodology and AI transparency pages.