A QA scorecard is only useful if two reviewers listening to the same call would give roughly the same score, and if the score tells the agent what to do next. Many scorecards fail both tests: they are long, vague and end in a number nobody acts on.
This article gives you a template you can copy into a spreadsheet, one version for service calls and one for sales calls, plus rules for scoring and a way to connect results to practice.
What a good scorecard does
- Measures behaviours you can hear. "Showed empathy" is hard to score. "Acknowledged the problem in their own words before offering a fix" is easy to score.
- Stays short. Five to eight criteria per call type is enough. Longer forms push reviewers toward ticking boxes.
- Separates must-haves from skills. Compliance items (identity check, required disclosures) are pass or fail. Soft skills are scored on a scale.
- Points to one change. The review ends with a single thing to try on the next call.
Template A: service calls
| # | Criterion | What you should hear | Score |
|---|---|---|---|
| 1 | Greeting and ownership | Name, company, offer to help; sets a clear start | 0-2 |
| 2 | Listening and understanding | Lets the caller finish; restates the issue accurately | 0-2 |
| 3 | Empathy | Acknowledges how the caller feels before moving to the fix | 0-2 |
| 4 | Accuracy | Correct information; says "let me check" instead of guessing | 0-2 |
| 5 | Clarity and tone | Plain words, steady pace, no jargon | 0-2 |
| 6 | Resolution or next step | Issue solved, or a named next step with a time | 0-2 |
| 7 | Close | Confirms what happens next; asks if anything else is needed | 0-2 |
Pass or fail items: identity verification where required, mandatory disclosures, correct data handling. A fail here overrides the score.
Template B: sales calls
| # | Criterion | What you should hear | Score |
|---|---|---|---|
| 1 | Opening | Who, why, and a reason for the person to stay on the line | 0-2 |
| 2 | Discovery | Open questions; the customer talks more than the seller | 0-2 |
| 3 | Fit and value | Links what was heard to what is offered, in the customer's terms | 0-2 |
| 4 | Objection handling | Acknowledges, asks a question, then responds | 0-2 |
| 5 | Control of the call | Summarises, moves the conversation forward without pushing | 0-2 |
| 6 | Next step | A specific agreed action with a date or time | 0-2 |
For examples of what each sales criterion looks like in practice, see Sales Call Scorecard: 5 Criteria to Score Every Call.
How to score consistently
- Define the three levels once. For each criterion write one line for 0 (not heard), 1 (partly), 2 (clearly heard). Keep the definitions beside the form.
- Score from the recording, not memory. Note the minute of each example you use.
- Calibrate monthly. Have two or three reviewers score the same call, then compare. Discuss differences until the definitions are clearer.
- Review a sample, not only the worst calls. If you only listen to complaints, agents learn QA is punishment.
- Give the agent the evidence. "At 2:10 you offered a fix before the caller finished" is more useful than "improve listening".
From score to practice
A score becomes useful when it turns into a practice goal. Pick the lowest-scoring criterion and make it the focus of a few short practice conversations.
For example:
- Low on empathy: practise the first thirty seconds with an upset customer.
- Low on objection handling: drill three or four objection replies out loud. See Objection Handling Practice.
- Low on next step: practise ending calls with a specific, dated agreement.
Practice works best in short, frequent sessions. A tool can help here. With DIRI AI you speak by voice with an AI customer in a prepared situation and read a report afterwards that points to moments to try again. It is practice, not a replacement for your own QA process: your reviewers still own the standard.
A one-page review routine
- Pick three calls per agent per month, including at least one good one.
- Score them with the template.
- In the conversation with the agent, start with what went well, then agree on one criterion to improve.
- Schedule two or three short practice sessions on that criterion.
- Re-score a new call next month and compare that criterion only.
Common mistakes
- Weighting everything equally. Decide which two or three criteria matter most for your business and weight them.
- Scoring tone without examples. Tone is real, but give the reviewer words to listen for.
- Changing the form every quarter. You cannot see progress if the yardstick keeps moving.
- Using scores to rank people publicly. It encourages gaming. Use the scorecard to coach.
FAQ
How many criteria should a QA scorecard have?
Five to eight per call type is a practical range. More than that makes reviews slow and scoring less consistent.
Should compliance items be scored on a scale?
No. Treat them as pass or fail so a missed requirement is never hidden inside a good average.
Can I use the same scorecard for sales and service?
Share the common parts (greeting, listening, clarity, close) but keep separate criteria for what is specific to each call type.
Related: Customer Service Phone Training, Call Centre KPIs (AHT, FCR) and How Training Affects Them, 12 Customer Service Roleplay Scenarios for Phone Training. Free tool: Sales Call Checklist.
Turn the lowest score into a practice call
DIRI AI has a Free plan: 3 voice calls a month, up to 5 minutes each, no card needed. Start on the Free plan and rehearse with an AI customer. Want a look first? Read the sample dialogues.