What do we measure?
We measure whether ChatGPT, Gemini and Perplexity mention your products when asked the questions your shoppers ask, how they describe them, and whether they recommend them. Each answer is recorded word for word, as the assistant gave it.
The result is a picture of your products' AI visibility: which questions bring them up, which do not, and where an answer about them is thin or wrong. That is where the app proposes a listing fix for you to approve.
Where do the questions come from?
The questions are written the way shoppers ask them, from two sources: search data about what people look for in your product's category, and your own products, starting from their own words and the answers you give in Intake.
The questions name products and brands, not people. The privacy policy names the search data provider and the AI assistant providers that receive them.
Why ask the same question more than once?
AI answers vary from one run to the next, even when the question is identical, so a single answer tells you little. We ask each question several times and report the range of what came back, not one lucky or unlucky answer.
Asking again is also how a real change is told apart from noise: a difference that holds across runs is worth acting on, and one that appears once may not be.
What does the range mean?
The range is the spread of results across runs: how often an assistant mentioned or recommended your product, from low to high. It describes what we saw, and it is not a promise of what the next answer will say.
When a new measurement lands inside the earlier range, the change may be noise. When it lands clearly outside it, something more likely moved.
What can the measurement not tell you?
It does not report sales, and it does not show what any one shopper sees, because assistants personalize their answers. It shows a point in time, the assistants change their models, and it covers only ChatGPT, Gemini and Perplexity.
Read it as a regular look at how assistants describe your products, next to your own sales and traffic figures, not as a replacement for them.
What happens to the answers we collect?
Every answer the app collects is kept, so your scores can be re-checked later against the exact words the assistants gave. The questions behind them name products and brands, not people, and the app reads no customer data to ask them.
What is kept after you uninstall, what is deleted and when, and who processes the data are set out in the privacy policy.
How do changes get checked?
Nothing is written to your store before you approve it. After each approved change, the app reads the product back to check that it landed, and records it in Activity with what it replaced, so support can revert it on request.
A later measurement then asks the questions again, so the answers before and after a change can be compared, each with its range. The full flow is on How it works.