Why one AI answer tells you nothing about your AI visibility
AI answers change from one run to the next. How to measure AI visibility properly: repeated questions, frequencies, ranges and the checks that avoid false findings.
Short answer: ask the same question many times and count. Report 'named in 3 of 10 answers' with a range, never a screenshot of one answer.
Answers vary a lot
When we asked the same question twice, the overlap between the two lists of brands was often low: between 0.24 and 0.70 on a 0-to-1 scale, depending on the question. Leading brands repeat; the rest of the list moves.
Count, with honest ranges
With 10 answers, '0 of 10' still allows a true rate of up to about 28% (95% interval). So the honest wording is 'rarely named', not 'never named'. More runs narrow the range.
Three checks before calling something a gap
- Is it the brand's real niche? A home-fragrance brand absent from perfume questions isn't a gap.
- Does it sell in that market? A brand that ships only in Europe being absent from US answers means nothing.
- Did the answer name any brand? Some answers give general advice with no brands; those aren't losses.
Check the apps too
We measure through official APIs, which let us repeat questions at scale. The leading brands match what the consumer apps show; the long tail can differ. Before a finding goes into a report, we check it in the app.