Questions, answered with the numbers.
Where do the prompts come from? Did you invent them?
Show the answerHide the answer
The queries are generated in the market's language from your website and shown to you before the first run. You edit any line or import your own from Search Console. The free snapshot lists its 10 queries before the first engine answers, so nothing runs behind your back.
Is this just a rank tracker for ChatGPT?
Show the answerHide the answer
No. There is no position to track, so we count how often your pages are cited as a source, per query, engine and country. The answer text sits behind every count. The sub-queries the engine ran show why being first on Google does not put you in the answer.
Do you show me the raw answers?
Show the answerHide the answer
Every cell opens the stored answer: the text, the cited pages in the engine's order, the sub-queries it ran and the brands it named. The free snapshot shows this for 10 queries on 3 engines; a paid workspace exports it all as CSV and JSON.
Can I edit the queries?
Show the answerHide the answer
In a workspace, yes: edit any line, pause one, mark it core, or import from Search Console. A measured query you edit becomes a new one, and the old one keeps its history. The free snapshot's 10 queries are fixed once the engines start.
What is "Why them"?
Show the answerHide the answer
The answer behind a competitor's citation: the stored answer the engine gave, the pages it cited with their paths, and the sub-queries it ran to get there. It opens from any cell in the report, because why is a competitor cited and not me is the question every number on the page raises.
Why do results change every time I ask ChatGPT?
Show the answerHide the answer
Your ChatGPT knows you: memory, history, location and account language shape the answer you see. The model itself can answer the same query 2 ways. We run from a clean session located in the market, 2 times a week on Growth. The cell prints a count such as 1 of 2, never 1 run as a verdict.
How many times do you run each query, and how noisy is it?
Show the answerHide the answer
1 run a week on Starter, 2 on Growth and Pro, plus the core queries daily. A single query moves a lot between runs, so we print counts of runs and a 95 % range, never 1 run as a percentage. A country score on Growth carries about ±4 points, a single query about ±35. That is why a query is called up or down only on 4 weeks of data.
Which engines do you cover in my country and language?
Show the answerHide the answer
Any country where an engine answers with a location. The matrix on the methodology page prints a cell per engine and country: available, partial or not available. Engines not available in a country are not billed. A country with 2 languages, such as Belgium in French and Dutch, is 2 markets with 2 query sets.
Are you asking ChatGPT, or the API? They are not the same thing.
Show the answerHide the answer
The API, from a fixed location in the country, in the market's language, with no browsing history, no memory and no signed-in account — the half of the answer that is reproducible and comparable week to week. A buyer with 3 years of chat history will see something else, and no tool on this market can measure that; anyone claiming otherwise is measuring an API too. What we do about it: the model id and the run number are printed on every cell, every query runs at least twice on paid plans, and the methodology page states the location, the model and the parameters for each engine.
Is this GEO?
Show the answerHide the answer
Some people call this generative engine optimisation. We call it measuring who the assistants cite for your buyers' queries in each country, and what to change. The name matters less than the count behind it.
Do I need a card, and what happens on day 15?
Show the answerHide the answer
No card, ever; on day 14 nothing is charged and nothing is deleted. The workspace becomes read-only and free and keeps every run and Playbook item. It keeps measuring 1 market on 10 queries and 3 engines every Monday, until you choose a plan.
Are there hidden fees, and can I cancel and take my data?
Show the answerHide the answer
The price on the card is the whole price excl. VAT: no setup fee, no per-seat charge, no minimum term on monthly billing. Cancel in Settings, effective at the end of the paid period. Export everything before or after: CSV on every paid plan, raw JSON on Growth and above. Data is kept 30 days after cancellation, then deleted.
How is this different from Semrush's AI toolkit or Ahrefs Brand Radar?
Show the answerHide the answer
3 things, all checkable. Every query is generated per country in its language and run from that country. Every number opens the stored answer with its run count, and a change you mark done is re-measured on the next run. The compare pages list what each tool states on its own site, dated, with the rows where the other tool wins.
What data do you keep about my brand and queries, and do you train on it?
Show the answerHide the answer
Your queries, the stored answers, your brand aliases, competitors and Playbook, on Render in Frankfurt (EU), with raw answers kept 12 months. No model of ours is trained on any of it. The engine APIs receive the query text only, on paid tiers whose terms exclude training on input; the subprocessor list links each provider's terms, and the DPA applies on sign-up.
10 queries, 1 country, no account.
The first engine answers within minutes. No account until you save the result. Then 50 queries a week, from €99.