Methodology

How we measure, in the open.

A number you can't check isn't worth reporting.

01 — The unit of measure

We measure conversations, not keywords

Persona→Topic→3 scenarios→3 turns→Score
02 — What we count

Three numbers that matter

illustrative
Recommendation rate

Mentioned isn't chosen

We report them separately. The loss sits between them.

illustrative
The engine's searches

What AI searches before it answers

You can't be cited by a search that isn't looking for you.

illustrative
Cited sources

Who writes what AI reads

That's where the work list comes from.

03 — Rules

Three rules written in code

A mention is not a recommendationBoth denominators, always printedThe average isn't the headline: we report per topic
04 — Limits

What we can't know

See the method applied

A full report (in Romanian), with the same rules and the same limits.