How we measure AI visibility

A single AI answer proves nothing: the same question can get a different answer tomorrow. That is why we measure repeatedly, document every run and follow fixed rules.

Which systems we measure

We measure seven AI systems: ChatGPT, Microsoft Copilot, Gemini, Google AI Overviews, Google AI Mode, Perplexity and Claude.

The Basic Check covers five of them: ChatGPT, Gemini, Google AI Overviews, Google AI Mode and Copilot. Copilot is included because 43% of AI users in Germany use it (Bitkom, April 2026).

How a measurement run works

  • We ask every question three times in every system.
  • Each run starts in a clean session with no history.
  • For every run we record: system, date and time, exact question, country and location method, interface language, logged in or not.
  • For DACH measurements we use a Germany-based environment wherever the system allows reliable location control. If a measurement is taken outside Germany, the report says so.

Baseline

Before any optimisation we measure the starting point. This baseline is dated and frozen. Every later result is compared against it as before → after.

Spread

AI answers vary from run to run. Three runs per question show whether a result is stable or appeared only once.

What we count separately

  • Mention: your company's name appears in the answer. If your company is only described or linked without its name, we count it as a soft mention.
  • Citation: the answer lists your website as a source.
  • Recommendation: the AI explicitly suggests your company as a provider.

A mention is not a recommendation. That is why we count all three separately.

How we label results

Every statement in our reports carries one of three labels:

  • Verified: measured directly, or backed by a source and date.
  • Inferred: our conclusion from the measurement data.
  • Open: not yet checked.

Limits of measurement

  • No measurement guarantees that a company will be mentioned, cited or recommended in future.
  • AI systems change constantly. Results apply to the period measured.
  • Location, language and login status affect answers. That is why we record them for every run.

Example: Basic Check

15 questions × 5 systems × 3 runs = 225 measurements. You receive a written report and a call to go through it. Price: €950 net, one-off. The amount is credited against a subsequent Full Analysis.

→ More about the visibility audit

Next step

Want to know what AI systems say about your company today?