How we measure
Counted, not claimed. Every number on the leaderboard is what the AI engines actually said when asked a real buyer question — a neutral, timestamped measurement, not our opinion or a paid placement.
The method
- Real buyer questions. For each category and city we ask the questions a prospect actually types — “best [category] in [city]”, “who should I hire”, “what does it cost” — never “what is [brand]”.
- Repeated, across engines. Each question is asked several times to ChatGPT, Claude, Perplexity and Gemini, because a single AI answer is noise.
- Frequencies with a confidence range. A brand’s visibility is the share of answers that name it. The ± is the 95% Wilson interval — two brands whose ranges overlap are a statistical tie, which the board shows as overlapping bands rather than a false ordinal gap.
- Same questions every month. So month-over-month movement is real, not a re-roll.
- Not gameable. We publish the method and citation receipts, but not the live prompt list.
Own a business here?
You can claim your listing to respond, flag an error, request a re-measure, and embed a rank badge — verified by a work email on your domain, or by a DNS record / homepage meta tag. Claiming never changes your number.
Want to climb? Find your category and run a free scan to see exactly which questions you’re losing.