Call center QA scorecard template

What a QA scorecard is for

A QA scorecard turns "that was a good call" into something an agent can act on and a manager can trend. It scores what the agent did on one interaction against the behaviors that drive resolution, compliance and customer experience. Done well, it is the main input to coaching. Done badly, it becomes a points game that agents resent and that tells leadership nothing.

Keep it separate from the monthly call center agent scorecard, which rolls QA together with handle time, adherence, attendance and survey results. The QA form judges one call. The agent scorecard judges a month. If you want the background on how quality programs are usually set up, our glossary entry on quality assurance covers sampling, roles and reporting.

How the weighting and auto-fail rules work

The template weights resolution and accuracy at 30 points and soft skills at 25 because that is roughly what moves customer outcomes. Opening at 10 points is deliberate: a perfect greeting on a call where the customer got the wrong answer should not score 85%.

Compliance items carry no points. They are pass/fail, and a fail sets the score to 0. That sounds harsh, but the alternative is worse: on a weighted points model, an agent can skip a required disclosure, lose 5 points and still pass with 92%, and your dashboard will say quality is fine while you carry real regulatory exposure. Keep the auto-fail list short (four to six items) and limited to things that create legal, financial or security risk.

Worked example: Denise takes a billing call. She earns 9 of 10 on the opening, 16 of 20 on discovery, 25 of 30 on resolution, 22 of 25 on soft skills and 12 of 15 on closing. Hold was not needed, so the 5-point hold item is N/A. Points earned = 84; points possible = 95; score = 84 / 95 x 100 = 88.4%, which lands in Meets. Had she skipped the recording disclosure, the score would be 0 and the conversation would be about that one miss.

How many calls to score and who scores them

A common benchmark is 4 to 8 evaluations per agent per month, more for agents in nesting or on a performance plan. Fewer than four and one bad call swings the monthly average too much. Many centers run one analyst per 40 to 60 agents, with supervisors scoring a few calls each week themselves so they stay close to the work.

If you use speech analytics to score 100% of calls automatically, keep human evaluations for the judgment items (empathy, accuracy, whether the resolution was right) and let the system flag silence, overtalk, missed disclosures and long holds. Automated scores are good at finding which calls to listen to. They are weaker at telling you whether the agent gave the right answer.

Calibration is what makes the numbers mean anything. Once a month, have every evaluator and two or three supervisors score the same three calls independently, then compare. A healthy target is evaluators landing within about 5 points of each other. If one analyst consistently scores 10 points lower than the others, agents on that analyst's list are being judged on a different standard, and they notice. See our guide to call calibration for how to run the session.

Common mistakes with call monitoring forms

• Too many line items. Scorecards with 30 or 40 behaviors produce averages in the high 90s and no clear coaching point. Twenty to twenty-five behaviors is plenty.
• Scoring scripts word for word. Unless it is a legal disclosure, score whether the agent achieved the purpose (verified the caller, set expectations), not whether they said the exact phrase.
• Only scoring short calls. Analysts under time pressure pick 3-minute calls. Long calls and transfers are where most problems sit, so stratify the sample by call length.
• Handing over a score without a coaching conversation. The form should end in a session on the call center coaching form with one behavior to work on, not five.
• Ignoring process failures. If half the misses on resolution trace back to a broken refund tool, that is an operations problem. Track the root-cause field and report it upward.
• Using QA scores alone to decide discipline. Use them as evidence alongside patterns, coaching history and other metrics.

Adapting the scorecard for multi-site and high-volume teams

When a program runs across several sites or vendors, use one master form with identical section weights everywhere and only localize the bracketed items. Report QA score by site, team and evaluator side by side; tools like site and zone performance reporting make the gaps obvious. If a new site averages 94% while its first-contact resolution trails the others, the evaluators are probably being lenient.

For new-hire classes in nesting, score more calls (two or three a day for the first week on the floor) on a shorter version of the form focused on verification, accuracy and hold handling. Trend the class average daily and look for the behavior most of the class is missing; that is a training gap, not an individual one. When you hire QA analysts, our quality analyst interview questions include a live call-scoring exercise worth stealing.

Record-keeping note: if you record calls, follow the call-recording consent rules for every state you take calls from; some states require all parties to consent. Rules vary by state and country, so check with counsel.