Methodology
How the data works.
How Searchtrack collects AI answers, and what its figures do and don’t mean.
What we collect
We ask ChatGPT, Gemini, Google AI Mode and Google AI Overviews five shopper questions in each of 500+ product categories, on a regular schedule. Questions cover the general “best to buy”, a common use, a budget, a common feature and brands. Answers are collected through data suppliers as logged-out, US English requests, so they are not personalised and may differ from what you see in your own account.
What a figure means
A brand’s share is the proportion of answers that name it. Leaderboards combine all four engines over recent scans, because a single engine on its own gives too few answers to rank reliably. Single-engine views show counts, not ranks, and a change is shown only when it is larger than the margin of error.
Evidence, not cause
Searchtrack records what AI answers say and which pages they cite. It does not know why a model chose a brand, and being cited by a source does not guarantee a recommendation. When we show events happening together, such as a new citation and a rise in share, we label them as happening together, never as cause and effect.
Estimates
Traffic and revenue figures on brand pages are estimates built from search traffic data, and are labelled as estimates. They are shown only for brands that have been enriched.
Brands and products
Brand and product names are read from the answers by code and checked by our team. Brand names and logos are used only to identify brands. Searchtrack is not affiliated with OpenAI, Google, Shopify or any brand shown.
Health and sensitive categories
For health, supplement and medical products we never show excerpts of AI answers that make health claims on public pages.
Corrections
If something about your brand looks wrong, email support@searchtrack.io and we’ll check it.