What we measure, and what we refuse to.

Search stopped being a list of links. When someone asks ChatGPT which supplier to use, one answer comes back, and either it names you or it does not. Vidima measures which of those two things is happening, on the prompts your buyers actually type, and how that changes week over week.

We track ChatGPT, Perplexity, Gemini and Google’s AI Overviews — the same prompts, week after week, in whatever language your buyers ask in rather than translated from English. The output is three things: an AI visibility score, your share of voice against named competitors, and a ranked list of fixes.

A single AI answer is close to a coin flip. Ask the same question twice and you can get two different lists of companies, because these systems are probabilistic and they re-read the web constantly. One answer is an anecdote; a run of answers over time is a measurement. That is the whole reason Vidima runs on a schedule instead of on a button.

• Prompts, not keywords. We track buyer-intent questions, in the buyer’s own language.
• Every answer is kept. A score always opens into the text that produced it — what the engine said, who it cited, and where you were missing. If you cannot click a number and read the answer behind it, do not trust the number.
• The same prompts, repeatedly. Movement is only meaningful against a stable set.

Vidima measures visibility inside AI-generated answers. It is not a blue-link rank tracker, and organic search positions are excluded from every score — including on Google, where only the AI Overview or AI Mode answer and the sources it cites are counted.

This means a business can rank first organically and still score zero here. That is not a bug in the measurement; it is the measurement. Ranking well in a list nobody is shown any more is exactly the gap this product exists to make visible.

• A scan is a sample, not a census. We ask a tracked set of prompts on a schedule. We do not see every question your buyers ask, and we never claim to.
• Traffic figures are a floor. If you connect Google Analytics we count sessions that still carried an AI assistant’s referrer. Copied links and in-app visits arrive with no referrer at all, and Google’s own AI surfaces arrive as google / organic, which Analytics cannot separate. The real number is higher than the one we show, and we cannot tell you by how much.
• Observed signals, not proven causes. When we say something appears to move an answer, we mean we observed it alongside a change. These engines are closed systems. Nobody outside them — us included — can prove causation, and anyone in this category who tells you otherwise is selling something.

We sell an AI visibility tool, and we publish research comparing AI visibility tools, including the ones we compete with. Those two facts sit badly together and we would rather say so than have you notice it yourself.

What we do about it: every comparison we publish states its method and its sample size, keeps the raw answers, and reports results that are unflattering to us alongside results that are not. Read them as evidence you can check rather than as a neutral verdict — the method is the part you should audit, not the conclusion.

Vidima is the trading name of Evolving Technologies EOOD (Еволвинг Технолоджис ЕООД), a company registered in Bulgaria and based in Sofia. The same entity is the data controller named in our privacy policy and the contracting party in our terms of service.

Questions about the product or the method: hello@vidima.ai. Anything about your data: privacy@vidima.ai.