A support QA scorecard is the weighted rubric a team uses to score customer-support interactions consistently, listing each quality criterion, how much it counts, and what qualifies as a pass, a fail, or an automatic fail.
A support QA scorecard is the weighted rubric a team uses to score customer-support interactions consistently, listing each quality criterion, how much it counts, and what qualifies as a pass, a fail, or an automatic fail. It is the instrument that turns a vague sense of “good support” into a repeatable number, and it is what makes QA scores comparable across agents, reviewers, and time.
What goes on a support QA scorecard
A scorecard groups the criteria that define a good interaction and assigns each a weight. A typical structure:
| Section | Example criteria | Weight |
|---|---|---|
Accuracy | Correct information, correct policy applied | High |
Resolution | Issue solved, first-contact resolution where possible | High |
Compliance | Verification, disclosures, regulated-industry rules | Auto-fail if missed |
Tone and empathy | Brand voice, sentiment read correctly | Medium |
Process | Escalation, documentation, follow-up | Medium |
The exact criteria and weights depend on the business. A bank weights compliance far more heavily than a game studio does, but the shape stays consistent.
Pass, fail, and auto-fail
Most scorecards produce a percentage score, but the more important design choice is the automatic fail: a small set of items so serious that missing one fails the whole interaction regardless of the rest, typically a compliance breach or a materially wrong answer. Auto-fails keep a high overall score from hiding a dangerous mistake. For example, an agent can handle a call warmly, resolve the issue, and still auto-fail for skipping identity verification on an account change, because that single miss carries more risk than everything done well.
How to build and use one
Keep it short enough to score quickly and consistently, give every criterion a named owner who will act on it, and calibrate it by having multiple reviewers score the same interactions until their results line up. A scorecard that takes too long to apply, or that no two reviewers score the same way, gets abandoned. For a full build walkthrough and template, see the support QA scorecards guide, and for the scoring process end to end, how to QA support tickets.
Why it matters
Without a shared scorecard, QA is just opinion, and opinions do not calibrate, trend, or coach. The scorecard is what lets a team compare this month to last, this agent to that one, and a human agent to an AI one, all on the same standard. As AI auto-scores more interactions, up to 100 percent of them rather than the 1 to 2 percent a manual team can reach, the scorecard also becomes the specification the model scores against, so a clear, well-weighted scorecard matters more, not less.1
See how unitQ compares on your data
A short demo, run on your own feedback.
Related terms
FAQ
QA every conversation against one standard
A scorecard only works when every interaction is measured against it. See how unitQ Support works to auto-score conversations at full coverage, not a sample.
Related guides
Sources 1 references
SQM Group, "Automated vs. Manual QA: How to Improve Accuracy, Insights, and Cost Efficiency." sqmgroup.com. Accessed October 2026.