Quality Assurance (QA) score is a composite rating of agent performance based on evaluated interactions, typically assessing accuracy, compliance, communication quality, and adherence to process.

QA programs exist because customer satisfaction scores alone don't tell you why interactions succeed or fail. A CSAT of 4/5 doesn't reveal that the agent gave incorrect policy information that will cause problems later. QA catches process failures before they compound into churn, complaints, or regulatory issues—critical for fintech and insurance where errors have legal consequences.

The challenge is calibration. Inconsistent scoring across QA reviewers makes the data useless. Establish clear rubrics, run regular calibration sessions, and track inter-rater reliability. Modern QA increasingly uses AI to score 100% of interactions rather than sampling 2-5%—this catches outliers that random sampling misses and removes the "performance on reviewed calls" vs. "performance on all calls" gap. Just ensure AI scoring is validated against human judgment before trusting it for coaching decisions.

Related terms: Agent quality score, Escalation rate, First reply time