When an AI model describes your brand, the tone matters. Sentiment analysis labels each mention as positive, neutral, or negative.
On a sliding scale of 0-100 the higher the score the more positive.
Sentiment is a valuable metric, because your brand can be highly visibile, but instead of being reccomended by LLMs its actively being warned against.
How sentiment is detected
For every AI response that mentions your brand, we extract the sentence(s) that reference you and run them through a sentiment classifier. The result is one label per mention.
- Positive — describes your brand favourably (“powerful”, “best-in-class”, “reliable”)
- Neutral — descriptive, no strong tone (“a SaaS company that tracks brand visibility”)
- Negative — critical or warning language (“limited features”, “expensive”, “buggy”)
What you’ll see in the dashboard
- Sentiment Score — Weighted average of recorded sentiment acrorss conversations.
- Sentiment trend — how the score shifts over time
- Sentiment prompt / chat drill-down — you can view sentiment down to a prompt or individual chat level.
Limitations
Sentiment is hard. Sarcasm, mixed reviews (“great UI but slow”), and tongue-in-cheek descriptions sometimes get misclassified. We err on the side of marking ambiguous cases as neutral.