A QA rubric, often called a quality scorecard, is the set of criteria used to score a customer service conversation and how much each one counts. It turns a vague idea of good service into line items reviewers grade the same way.
In short
- Good criteria are visible in the transcript, so two reviewers agree.
- Weighting makes critical items count more than minor ones.
- An auto-fail item, like a compliance breach, can zero the whole score.
- The less subjective the rubric, the more defensible the scores.
Common QA rubric criteria
A QA rubric groups its line items into the areas that define a good interaction. Typical criteria include:
- Accuracy: the agent gave correct and complete information.
- Process adherence: required steps, identity verification and disclosures were followed.
- Tone and empathy: the agent was clear, professional and understanding.
- Resolution: the customer’s issue was actually solved.
- Efficiency: the issue was handled without unnecessary friction or delay.
Each criterion should be phrased so it can be answered from evidence in the conversation, not from a reviewer’s gut feel. For a full worked example of a scorecard, see our guide to the QA scorecard.
How weighting works
Not every criterion matters equally, so a rubric assigns weights. A wrong answer that misleads the customer should cost more than a small tone slip. Weighting makes that priority explicit in the score.
| Criterion | Example weight | Why |
|---|---|---|
| Accuracy of information | 30% | A wrong answer directly harms the customer |
| Process and compliance | 25% | Missed steps create real risk |
| Resolution | 25% | The customer’s problem is the point |
| Tone and empathy | 20% | Shapes experience but rarely breaks it |
Auto-fails and consistency
Some misses are severe enough that no amount of good behavior should offset them. A compliance breach or a rude interaction can be set as an auto-fail that zeros the score regardless of the other criteria. Used sparingly, auto-fails keep the rubric honest about what is non-negotiable.
A rubric is only as good as its consistency. When criteria are subjective, reviewers disagree and scores lose their meaning, which is why teams run calibration. Kaizo applies your rubric to 100% of conversations automatically, with every score linked to the evidence in the transcript, so the same standard lands on every interaction without reviewer drift. At UiPath, Kaizo automated 100% of QA with 200% ROI, applying one consistent rubric across every conversation.
Frequently asked questions
What is the difference between a QA rubric and a QA scorecard?
They refer to the same thing: the structured set of weighted criteria used to grade a conversation. Rubric emphasizes the grading logic, while scorecard emphasizes the tool reviewers fill in. Our QA scorecard guide walks through a complete example.
How many criteria should a QA rubric have?
Enough to capture what matters and no more. Most effective rubrics keep to a focused set of clear, weighted criteria, because a long list of vague items is harder to score consistently and slower to review.
What makes a good QA rubric criterion?
It is observable and objective: a reviewer can point to the moment in the transcript that proves whether it was met. If a criterion cannot be tied to evidence, different reviewers will grade it differently and the score becomes unreliable.
How does weighting affect the score?
Weighting decides how much each criterion contributes to the total, so high-impact items like accuracy and compliance move the score more than minor ones. Auto-fail items go further and can zero the score outright for a serious miss.
