Quality
How every conversation scored against each quality parameter, and where it slipped.
Parameters, weights, scoring instructions, auto-fail rules and sampling are set in Console › Quality scorecards. Scores here are produced by those instructions.
Conversations scored
742
100% AI scored
Weighted quality score
81%
Target 85%
Auto-fails
23
↑ 6 vs last week
Human review coverage
4.2%
31 conversations
Support v4 — score by parameter
Weight, average score and how often the parameter was missed
| Parameter | Weight | Score | Missed | Scoring instruction | |
|---|---|---|---|---|---|
| Greeting and identification | 10% | 93% ↑ 2 pts vs last period | 31 of 412 | Agent states name and verifies two identifiers | |
| Discovery | 20% | 81% ↓ 3 pts vs last period | 78 of 412 | Asks at least two clarifying questions before proposing a fix | |
| Policy accuracy | 25% | 74% ↓ 6 pts vs last period | 107 of 412 | Quotes the current policy without inventing terms | |
| Empathy | 15% | 88% ↑ 1 pts vs last period | 49 of 412 | Acknowledges frustration each time it is expressed | |
| Resolution | 20% | 86% → 0 pts vs last period | 58 of 412 | States what happens next with a timeline | |
| Closing | 10% | 69% ↓ 8 pts vs last period | 128 of 412 | Recaps the case number and thanks the customer |
Score by interface
Click a channel to see its scored conversations
Agents below target
Weighted score against 85% target
Team average84%
Target 85%
Auto-fail triggers
Any match scores the conversation zero
- No recording disclosure
- 14 conversations
- Card details read aloud
- 5 conversations
- Promise outside policy
- 3 conversations
- Rude or dismissive language
- 1 conversation
Where quality is configured
Jump straight to the setup that drives these numbers