A call center quality review routine

Choose a sample that answers a question

The QA packet should include a sampling log, review card, evidence note, coaching action, and recheck date. Review card items need a definition, observable pass condition, not-applicable option, and weight only where management has justified it.

Create an evidence-based review card

Log interaction ID, call type, agent, reviewer, selection reason, review date, verification result, greeting, intent capture, required fields, authority compliance, routing, promise accuracy, note quality, outcome, evidence timestamp, defect type, action owner, and recheck result.

Score behavior, not personality

Sampling can rotate among random calls, targeted risk calls, new workflows, and follow-up checks. State which question each sample answers so managers do not generalize beyond the evidence.

Protect reviewer authority

Reviewers may score against the published rubric and recommend coaching or process repair. They may not infer motive, rewrite evidence, penalize behavior absent from the rubric, expose recordings broadly, or make employment decisions outside their role.

Resolve disputed and inaccessible calls

Missing or corrupted recordings should not become automatic failures. Material rubric ambiguity goes to calibration; privacy concerns use restricted review; disputed scores require a second reviewer and preserved original score.

Calibrate before coaching

Double-score a routine call, a multi-issue call, a call with no recording but a note, and a boundary violation with a polite tone. Discuss evidence differences before reviewing the wider team.

Track whether fixes hold

Monitor sample mix, agreement between reviewers, defects by type, not-applicable use, coaching completion, repeat defects at recheck, and process changes created. Avoid a single blended score that hides critical boundary failures.