Our methodology — how the AI Visibility Score is calculated
A score is only worth what its recipe is worth. Tools in this category hide their kitchen; here is ours, in full.
1. Realistic questions, approved by you
For each brand, RadarIA generates 15 questions the way real buyers ask them to AIs (“best X”, comparisons, budget, reviews, alternatives). You review, edit and approve them — YOUR questions get measured, not ours.
2. 8 AIs, genuinely queried, every day
Every day, each question is asked to ChatGPT, Claude, Gemini, Perplexity, DeepSeek, Mistral, Google AI Overviews and Google AI Mode — through their official APIs (for Google, which has no API for its AI answers: the results page, collected by DataForSEO), no staging. An AI that is down one day doesn’t invalidate the measure: it is simply marked absent.
Google AI Overviews (the AI summary at the top of results) and Google AI Mode are read from the Google results page, collected by DataForSEO for the brand’s country and language (and its city from the Agency plan, when DataForSEO knows it). Google does not show an AI answer for every question: “no AI answer” is recorded separately, outside the score — neither “not cited” nor down. A collection error, on the other hand, counts as down, as for any AI.
3. Faithful citation extraction
In every answer, we detect which brands are cited, at what rank, and in what context. The AI Visibility Score is the percentage of question × AI combinations where your brand appears, over the last 7 days.
4. What a score means — and doesn’t mean
AI answers vary from one query to the next: a score measures a sample at a point in time, not an absolute truth. We commit to the method — realistic questions, genuinely queried AIs, faithful extraction — never to a score level. A tool that promises you a number is lying to you.