Methodology

Per search term we ask 12 questions the way a buyer would, to ChatGPT, Claude, Perplexity and Gemini. That is 48 measurements, all with web access on, so you see what a customer gets today.

We check every answer for your business name and its known spelling variants. At the same time we log which competitors the assistant recommends instead.

Your score runs from 0 to 100. Per assistant, that is the percentage of questions where your business gets named. The total score is the average across the four assistants.

An outage at an AI provider does not count against you: unanswered questions fall out of the calculation. Your score never drops because of someone else's downtime.

AI assistants do not give the same answer every time. A single measurement is a snapshot, not a guarantee.

That is why we repeat the same 48 questions every week. You see a trend instead of a coincidence, and you know whether a fix actually works.

Alongside the AI questions, we check your own website: are your prices, hours and other core facts readable in the text and in structured data (JSON-LD)?

This part uses no AI. It is a direct check on your source code, so for every fact we find you get the page path and the literal text fragment as proof.

We report only what we measure: mentions, sources and facts from your source code. No invented numbers, no sentiment score we made up.

We measure your position again every week. Nobody can guarantee a fixed spot in ChatGPT, because an assistant builds every answer fresh.