# Build a QA scorecard for support conversations

A rubric to review and coach customer interactions consistently.

- Topic: Customer Support
- Framework: RCTFVV
- Level: Advanced
- Tags: Legal, Customer Support, Analyze, Fill-in Template
- Source: https://www.thinkwithprompt.com/prompts/build-a-qa-scorecard-for-support-conversations

## Prompt

```text
Role: Act as a quality analyst designing the support QA scorecard.

Context: Channels covered: [CHANNELS]. What good service means here (resolution, accuracy, empathy, efficiency, compliance, brand): [QUALITY_DIMENSIONS]. Compliance / policy must-dos and hard fails: [COMPLIANCE]. How the score is used (coaching, not punitive; any thresholds): [USE]. Volume and who does reviews: [REVIEW_CONTEXT].

Task: Produce: (1) the scorecard - criteria grouped by dimension, each weighted, with a 3- or 4-point behavioural anchor (what earns each score, with an example); (2) auto-fail rules (compliance/safety breaches) and how they interact with the score; (3) a 'not applicable' rule; (4) calibration guidance so two reviewers score similarly, including 2 sample interactions scored; (5) how the results should feed coaching, not just a number.

Format: Sections: Scorecard (table: criterion | weight | anchors) | Auto-fail rules | N/A handling | Calibration guidance + 2 scored samples | Coaching use.

Verification: Flag anchors that are vague ('was empathetic') rather than behavioural; ensure the weights sum correctly and no criterion double-counts another.

Validation: Have two people score one sample with the anchors: do they land within one point? If not, tighten the anchor that caused the gap.
```

## What to fill in

- CHANNELS
- QUALITY_DIMENSIONS
- COMPLIANCE
- USE
- REVIEW_CONTEXT

Free to copy and use. From ThinkWithPrompt (https://www.thinkwithprompt.com/), a library of 2,663 curated AI prompts.