Answer quality and resolution
Every conversation gets a 1 to 5 score for Dazza’s answers, with a reason, and whether the customer’s need was met.
Dazza judges every conversation, so you can see how well it’s doing without reading them all.
What’s judged
When Dazza labels a conversation with its topics, sentiment and language, a minute after it goes quiet, it also records:
- Need met: yes, no or unclear. Did the customer get what they needed, from Dazza or from your team? Unclear means they went quiet before it was settled.
- Quality of Dazza’s part: 1 to 5, for Dazza’s own replies only, never your team’s.
- A reason: one line, such as “answered from the returns page” or “guessed a delivery date”.
| Score | Means |
|---|---|
| 5 | Accurate, from your knowledge, and complete. |
| 4 | Right, but could be clearer or shorter. |
| 3 | Partly helpful, or honestly said it didn’t know and offered the team. |
| 2 | Vague, unhelpful or missed the question. |
| 1 | Wrong, made something up, or promised something it can’t. |
Conversations where Dazza didn’t reply have no score. If the customer writes again, the conversation is judged again.
In the inbox
The side panel shows Quality: whether the need was met, the score and the reason. On smaller screens they show under the customer’s name. Relabel judges it again.
In Insights
Answer quality shows, for the period and channel you’ve chosen:
- the average score of Dazza’s answers
- how often the customer’s need was met, with the split between yes, unclear and no
- the lowest-scoring recent conversations, with the reason, each linking to the conversation
Start with the lowest scores. If Dazza guessed or got it wrong, correct the answer or add what was missing to your knowledge.