User Sentiment
How people feel during and after their conversations.
The User Sentiment card shows a breakdown of the emotional tone detected across analyzed conversations, based on what people actually expressed — not an inference about how they might feel.
What you see
A donut chart split into three segments:
| Sentiment | What it means |
|---|---|
| Positive | Thanks, appreciation, satisfaction, excitement — "great", "perfect", "awesome". |
| Neutral | A matter-of-fact exchange with no emotional signal either way. This is the default for most conversations. |
| Negative | The user showed frustration, disappointment, or made a complaint. |
The headline number is a "satisfactory" rate, calculated as positive plus neutral over the total — neutral conversations count toward it, since they aren't a bad sign, just an unremarkable one. Only genuinely negative sentiment pulls the rate down.
How to use this
- Watch the Negative share over time more closely than the absolute Positive count — a rising negative share is the earliest warning sign of a problem.
- Cross-reference with Resolution Rate — negative sentiment with a high resolution rate can mean people are getting answers but not enjoying the process (tone, pacing, or friction issues).
- If negative sentiment clusters around certain topics, check Top Keywords and Failure Analysis to find the pattern.
Last updated on