Our method, in the open.
AI visibility is measurable — but only with a clean setup. Answers vary between runs, between days and between languages. Anyone showing you a clean percentage without a range either measured once or is rounding in your favour. Here is how we work instead.
A fixed query set
50 to 100 questions from your sales reality, defined together with your sales team, then measured unchanged. Only a frozen set produces comparable time series. If the set changes, the series restarts — and the report says so.
Repeated measurement with stated ranges
Every question several times, across several days, English and German separately. We report ranges, not point values. The range is not a weakness of the measurement; it is a property of what is being measured.
Three systems, because your customers use three
ChatGPT, Perplexity and Google (AI Overviews and AI Mode) cite differently and from different sources. A provider measuring only ChatGPT is measuring a third of your market.
Two modes, always declared
Mode A is live measurement in the browser with original screenshots. Mode B is source-based analysis, used where live access is not possible or not reproducible. Every report names its mode in the method box. We never mix the two silently.
A finding is an existence proof, not a rate
If we show that AI sent a buyer to your competitor three times, that means it happens. It does not mean it happens in some particular share of all cases. We write that distinction into every report, even where it makes our own number smaller.
What we do not do
No bought reviews, no invented quotes, no commissioned comparison lists, no manipulation of third-party content. Contractually committed. And we tell you openly what does not work — llms.txt and schema markup as supposed AI levers, for instance.
Tooling and data
Monitoring on EU-hosted tooling. Data processing agreement available before the first conversation. Your raw data is yours and is deleted on request; our anonymised industry benchmarks contain nothing that identifies your company.
See the method applied to a real case.
The free check applies exactly this setup to your company — at small scale, with the same method box.
Request the free check