We measure, we study, we implement
Every measurement is 100 recorded observations across five providers, then the implementation work and a fresh measurement. We talk about what the data shows, not about what sounds plausible.
We measure how visible companies, services and projects are in ChatGPT, Claude, Gemini, Grok and Perplexity, how their competition behaves and what actually changes model answers. Our measurement data is anonymized, so we speak about patterns and dependencies, never about a single company. It is material for an article, a book, an interview or a talk.
Why us
Every measurement is 100 recorded observations across five providers, then the implementation work and a fresh measurement. We talk about what the data shows, not about what sounds plausible.
We share no client results and no client names, in any form. We talk about patterns across the whole body of measurements, and about the convergences and dependencies visible in it.
Model answers are not deterministic, and absence of evidence is not evidence of absence. We always state the limits, including in a conversation with a journalist.
What we can talk about
Four things newsrooms, book authors and event organizers come to us for.
Individual company results stay with their owners. For media we prepare the aggregate picture: distributions, repeatable patterns and dependencies visible across the whole body of measurements.
Every figure we give has a source and a date. The whole measurement method is published: the number of questions, the providers, the index weights and the limits of inference.
For authors and newsrooms we explain the mechanism: how models pick companies, what changes a recommendation and which dependencies show up between an action and a result.
We give talks and join panels on AI visibility: for newsrooms, industry conferences and marketing events.
We do not share client data or client names. Market figures come with their source, and patterns from our own measurements only in aggregate, anonymized form.
How it works
The topic, the format and the deadline. If you are on deadline, say so in the first line.
Aggregate figures with their sources go out by email. For an interview or a talk we propose specific slots.
You can send a passage for a factual check before publication. We do not touch the argument of the piece, only the figures.
Who it is for
Writing about search, AI and marketing, and needing figures that can be verified.
Describing the shift in search and needing the mechanism, not just the headlines.
Looking for a guest who talks about measurement specifically and does not sell from the stage.
Wanting to understand the methodology before citing a result in their own report.
Media contact
We reply by email within one business day. Name the figure or the pattern you need, and flag a same day deadline.
Anonymized data and comment for media are free of charge. We only ask that you credit the source when quoting figures.
We check figures if you ask. We do not touch the content or the conclusions of your piece.
If we have not measured something, or the data is inconclusive, we say so instead of handing you a convenient quote.
Tell us what you need and by when. If we have measured it, we send the answer in aggregate form, with the source and the method. If we have not, we say so straight away.
Common questions
No, and there is no exception to that. Individual company results stay with them; for media we show aggregate, anonymized patterns only.
We comment on the mechanism and the pattern, not on one company's result. If you ask about a publicly described case, we speak to what is visible from the outside.
No. Media material is free. We only ask that you credit the source when quoting figures.
Within one business day. Put a same day deadline in the subject line.
It depends on the format and the venue. For industry and newsroom events usually not; for closed workshops we agree it separately.
Polish and English, both fluently, including live.