Methodology
How we measure whether AI recommends you
Here's what we ask, which engines we ask, how many times, and what we record, so you can verify every result.
Real buyer questions
We build the questions your customer would ask in their language and market. The answers suggest new questions, and we run a second round with them.
Multiple samples, multiple engines, recorded sources
Each question is asked twice on ChatGPT, Perplexity, Gemini and Grok. We record which businesses appear and the cited sources. When your name appears, we detect it by searching for it literally in the answer, not by asking another AI to judge.
Verified competitors
Competitors come from what AI actually recommends. We check that they do what you do, verify their domain by visiting their site, and crawl it. A feature most of them have is table stakes; one almost nobody has is an opportunity.
A number you can defend
In the site scan, each issue counts once even if it repeats across many pages, so crawling more pages does not sink your score. Anything we could not measure shows as "not measured", never as a pass.
Limits
AI answers change over time and depending on who asks. Structured data helps engines understand you, but it doesn't guarantee citations. We don't promise positions or recommendations.
Frequently asked questions
What is the AXO test?
An AI agent tries to complete a real task on your site, like getting a quote, booking or buying, and we record where it gets stuck. It is in development; today the scan checks the technical signals that make it possible.
Can I re-measure after making changes?
Yes. We use the same questions and engines so the comparison against your baseline is fair.