When an AI model describes your brand, the tone matters. Sentiment analysis labels each mention as positive, neutral, or negative.
Sentiment is scored 0–100 — the higher, the more positive.
Sentiment is a valuable metric, because your brand can be highly visible, but instead of being recommended by LLMs it’s 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 across 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.