Call Center Quality Assurance

How call center QA works

A standard QA program has four moving parts.

1. The standard. A scorecard that defines what a good interaction looks like for each channel and line of business. Our call center QA scorecard is a ready starting point.
2. Monitoring. Quality analysts or team leaders review recorded calls, chats and emails, and sometimes listen live. See call monitoring for the methods.
3. Calibration. Analysts, team leaders and sometimes the client score the same contacts independently, then compare and agree, so a score means the same thing no matter who gave it. More on this under call calibration.
4. Feedback and action. Scores go to the agent with specific coaching, trends go to training and operations, and repeated process failures go to whoever owns the process.

The point is improvement, not a score. A QA team that produces a lot of evaluations and very little change on the floor is an expensive audit function.

What goes on a QA scorecard

Most scorecards have five to eight sections and 15 to 30 items. A typical structure:

• Opening: greeting, name, ownership statement.
• Verification and compliance: identity check, required disclosures, call recording notice, payment card handling. Usually the critical items.
• Discovery: questions to understand the real issue, confirming the need.
• Resolution: correct information, correct process followed, issue solved or correctly escalated, accurate notes in the CRM.
• Customer experience: tone, empathy, clarity, managing holds and transfers, no dead air.
• Close: confirming resolution, setting expectations for any follow-up, closing politely.

Keep items observable. "Was empathetic" invites argument. "Acknowledged the customer's problem before moving to the solution" can be scored consistently. Weight the items to match what matters: compliance and resolution should carry more points than script phrases. The QA score then feeds into the wider agent scorecard alongside handle time, adherence and customer satisfaction.

How many contacts to score

A common practice is 4 to 8 evaluations per agent per month for tenured agents, and more for new agents, often 2 to 5 per week during nesting and the first month on the floor. That gives each agent enough feedback to act on without overwhelming QA.

Sampling is the real weakness. Five calls out of 1,200 a month is under half a percent of an agent's work, and random selection misses the calls that matter most. Better programs target the sample: long calls, repeat callers, transfers, low survey scores, and specific call types tied to complaints. Speech and text analytics tools now let centers apply automated checks to every interaction, such as whether a disclosure was read, and use human analysts for judgment-heavy reviews. A practical staffing ratio for manual QA is one analyst per 30 to 60 agents, depending on contact length and how many evaluations each agent gets.

Benchmarks and what a good QA score is

Targets are set internally, so the number depends on how strict the scorecard is. Many centers set the bar between 85% and 90%, with critical items required to pass. A center where almost every agent scores 95% or more usually has a scorecard that is too easy or analysts who are not calibrated, and a center where the average is 70% usually has a scorecard nobody believes in.

More useful than the average QA score is how it lines up with outcomes. If agents with high QA scores do not also have better customer satisfaction and resolution, the scorecard is measuring the wrong things.

Common QA mistakes

• Scoring scripts, not outcomes. Points for saying the exact closing phrase while the customer's issue went unsolved teaches agents to recite.
• Feedback that arrives two weeks later. Agents do not remember the call. Aim for feedback within two or three days, the same day for critical failures.
• No calibration. Two analysts scoring the same call 15 points apart destroys agent trust in QA.
• QA used only for discipline. If evaluations mostly appear in performance plans, agents dread them and stop learning from them.
• Ignoring process failures. If 30% of failed calls trace to a confusing refund policy, coaching agents will not fix it. QA should report root causes up as well as down.

Making QA improve performance

Link QA to coaching. Each evaluation should end in one or two specific behaviors for the agent to change, with a follow-up check on the next few calls. Let agents listen to their own calls and self-score sometimes; it shifts the conversation from defending a score to spotting the gap. Share strong calls as examples, not only weak ones. And feed QA data back into hiring and training: if new hires keep failing the same item, the classroom or the assessment needs to change, not just the agent. The call center quality analyst interview questions help when you are hiring for the role itself.