How it works

We ask the four big AI models the kinds of questions your shoppers actually type, then we measure where you show up. Here's the short version.

1. We ask the major AI models

   ChatGPT, Claude, Gemini, and Perplexity each get the same prompts — the questions a real person in your category would ask. Five repeats per question, so we're not just catching one lucky run.

2. We measure mentions, citations, position, and sentiment

   For every response, we check whether your name turns up, where in the list it lands, what tone the answer takes about you, and whether the model linked back to your domain. Citation and mention are different signals; we score them separately.

3. We score each model and aggregate

   Each provider gets its own 0–100 number, then we roll them up into one headline score with a confidence interval. Per-provider breakdowns stay in the report so you can see who already knows you and who doesn't.

4. You get a report you can act on

   Not a dashboard full of vibes — a real list of where you show up, where competitors edge you out, and which prompts are worth working on first. Every claim links back to the raw response.

Every prompt, model version, and rubric line is published on our methodology page. If something in your score looks weird, you can trace it.