Sentiment is the rating an AI answer attaches to the mention of a brand: recommended, listed neutrally, mentioned with reservations or advised against. It is the second dimension of every mention. The Mention Rate says how often a brand appears; the sentiment says whether that is good. The principle: mentioned is not recommended.
How does sentiment arise in AI answers?
A language model adopts the rating from its sources. For questions like “Is provider X good?” or “What experiences are there with Y?”, it draws on review platforms, forum threads (community presence), test reports and trade articles, and synthesises an assessment from them. Contradictory sources lead to phrasings like “mixed experiences”; uniformly positive sources lead to a recommendation; a single prominent negative review can shape an answer if it sits in a heavily weighted source.
Correctness is also part of sentiment in the broader sense: if the system names outdated services, wrong locations or a former positioning, a false picture emerges even if the tone is positive.
Why does sentiment matter for AI visibility?
Because trust and reputation questions come immediately before the purchase decision. A user who already has a brand in mind validates it via the AI, and whoever fails to convince here has lost a user they already had. A high Mention Rate with negative sentiment is worse than no mention.
What does this mean for your website?
Collect the sentiment per prompt category, especially for trust questions. Work where it arises: current reviews on the platforms that are cited; factual responses to critical voices; presence in forums with substance instead of advertising; a consistent brand description everywhere, so the system does not adopt an outdated version. And answer the question “Is a mention always valuable?” honestly: where do you want to be mentioned, and where not?