(tone) assesses how a brand is described in an AI answer. Examples range from positive ("particularly suitable for SMEs") through neutral ("is a provider of marketing automation") and mixed ("strong in features but complex") to negative ("is often criticized for poor support").
What matters is not the single phrasing but the recurring pattern across many prompts and platforms. Key questions are:
- Which source supports the statement?
- Is it current or outdated?
- Does it repeat across several platforms?
- Does it concern a real product or service issue?
Negative sentiment often arises from old reviews, third-party sites, support problems or outdated information. It cannot be controlled directly, but it can be softened over time through genuine root-cause fixes, current and consistent counter-evidence, and well-maintained third-party profiles.
Why sentiment shifts matter #
AI assistants compress many sources into one answer. If reviews, forum threads or news coverage turn negative, that tone propagates into AI answers quickly and at scale — often before a team notices it in classic channels. A sentiment drop across AI answers is an early warning signal for reputation issues.
Working with sentiment data #
Track sentiment per topic and per assistant, not just globally: a brand can be recommended for one use case and warned about for another. When sentiment dips, trace the sources the assistants cite — fixing the underlying coverage (support issues, outdated reviews, missing counter-narratives) is what moves the metric back.