Presence in AI responses
Pick a sector. We show the questions we measured this month and the sites AI cited to answer them. It’s the real measurement, done days ago — not a simulation. No cost.
We use your email to record this test. It does not authorise marketing messages. We delete this test record after 90 days. Privacy notice for the tests
Pick a sector on the left. Here appear the questions measured this month and the sites AI cited as source to answer each one.
How it works
A fixed set of questions is what makes the measurement comparable. Changing questions mid-series breaks it, so they are fixed with you at the start and only change on request.
Step 1
The questions your customers would ask an assistant before choosing. Written once, with you.
Step 2
ChatGPT, Perplexity and Gemini, all with web search enabled, plus Google’s AI Overviews. We record who is named, which competitors appear and which sites are cited as source.
Step 3
We cross-check with what your site publishes for machines and tell you the three things to do next month. The history stays.
What this is, and what it is not
Many sell AI presence without saying what they measured. Here, in writing, is what’s in and what’s out.
The models’ responses with web search, requested via API, and Google’s real AI Overviews for each question.
Each person’s ChatGPT responds with their history and memories. Nobody measures that from outside, and we don’t pretend we do.
Half the questions we measure bring none. When none appear, the report says «Google did not serve overview» — never «none appear».
Perplexity and Gemini almost always do; ChatGPT almost never. Where there are no sources, the report says there are none.
Nobody controls what a model answers. We promise to measure, explain why and fix what’s on your side.
Each month adds a point to the series. This is what allows you to see whether what you did worked.
In two lines
We measure model responses using web searches requested via API, and the actual Google AI Overviews. This is not the ChatGPT app with your session logged in, and we do not promise to get you there — we promise to measure and explain.
Price
Includes the 20 questions, the four sources, the monthly one-page report for you, and the history. You can cancel whenever you want; the history stays with you.
Do you already know your site is not being read as it should be? Correcting what your site publishes for machines is another service, on-demand: what AI reads from your site.
Questions
We do not promise that. What we do is measure, every month, whether you appear or not, who appears in your place, and which sites the AI is using as a source. Then we fix what is on your side. Anyone promising the rest is selling you something.
Because the measurement only makes sense when compared to itself. If the questions change every month, the series becomes meaningless. They change only at your request, and it is recorded when they do.
No. SEO measures positions on search engine results pages. This measures what an assistant responds — which is text — and who is named within that text.
For measurement, no. We only need your company name and domain. Access is only required if you want us to make the corrections.
The first measurement is immediate and is your starting point. Changes in model responses take weeks to months, and there are months when nothing changes. We tell you when that happens.
Start
Leave your email and company. We’ll reply with the 20 proposed questions for your sector and the first measurement. No calls.