How we measure AI visibility

Each prompt × engine × country combination runs weekly on Starter and more often on higher plans (up to daily) rather than once. AI answer engines do not return the same answer every time you ask — the same prompt run repeatedly can surface a different list of brands on different runs.

Every metric Leadee reports carries a 95% confidence interval computed with the Wilson score method, the standard approach for proportions with a small number of trials. A change is only flagged as significant when the confidence intervals of two periods no longer overlap — that is the signal that something real moved, not just normal variation.

30+ open prompts per category, run monthly through official AI engine APIs (OpenAI web search, Google AI Mode via DataForSEO, Perplexity, Gemini grounding). Each metric carries a Wilson 95% confidence interval, and a change between two months is only flagged when it is statistically significant. Brand names are the ones AI answers actually mentioned, normalized by a reviewed alias list; a newly seen name queues for approval before it appears. This ranks how often AI answers mention a brand for these prompts — it is not an endorsement or a quality rating.

The public AI Visibility Index applies this same measurement discipline — Wilson confidence intervals, official APIs only, significant-change flagging — to a fixed public prompt panel per category, published monthly.