Quality/Scorecards
Quality Scorecards
Weighted scorecards by use case, industry and agent — versioned with approval workflow, critical metrics and auto-fail rules.
Template weights
Example production scorecard definition
Outcome completion: 25%Factual accuracy: 15%Workflow adherence: 15%Customer experience: 15%Latency: 10%Speech quality: 10%Compliance: 10%Min pass 80Critical: compliance, hallucination
Weekly QA — Collections
Collections Agent · Jul 28 – Aug 3 · n=420 · v2.1.0
91
Task Success94 · w30%
Accuracy92 · w25%
Compliance96 · w25%
CX Tone88 · w20%
pass (min 80)trend uppercentile p72Collections
Improvement: focus on lowest weighted section; review evidence links on failed evaluators.
Weekly QA — Support
Support Concierge · Jul 28 – Aug 3 · n=380 · v2.1.0
82
Task Success84 · w30%
Accuracy80 · w25%
Compliance90 · w20%
CX Tone78 · w25%
pass (min 80)trend downpercentile p72Support
Improvement: focus on lowest weighted section; review evidence links on failed evaluators.