What we record
The exact question, target location or market, language, assistant, available model and search settings, date, saved answer and visible sources. Named-business checks are kept separate from discovery questions. We record the number of repetitions and keep unfavourable observations.
What we compare
Whether a business was mentioned, which other businesses appeared, whether its site was linked, whether the facts were accurate and whether the next step worked. Rechecks use the agreed question set, with any changed conditions noted.
What the evidence does not establish
A mention is not a customer. An observed answer is not a permanent ranking. A cited source does not prove what caused a recommendation. Enquiry and referral evidence is reported separately, with attribution limitations.
How we use consumer research
Each published statistic retains its source, geography, survey period, population, tool scope and limitations. Changing the target market in your link does not change where a study was conducted. These local-business surveys do not establish how online-only shoppers discover products.
We do not currently publish a verified global percentage of salon clients, clinic patients or restaurant guests acquired through ChatGPT. Owner interviews help us understand concerns; a single interview cannot establish a customer percentage.