How the QA score is calculated
QA score = points earned ÷ points that apply × 100. "Yes" earns the full points, "Partly" earns half, "No" earns none, and "N/A" lines are removed from both totals, so an email is not penalised for a hold-time rule. If any auto-fail item is answered "No", the score is 0 whatever the rest: some mistakes, like a wrong answer or sharing account details without checking identity, outweigh good manners.
Example: one phone call
With the phone scorecard above (100 points), an agent who meets every line except "Confirmed the issue back" (No, 5 points) and "Showed empathy" (Partly, 2.5 of 5 points) earns 92.5 of 100 points: a QA score of 92.5, excellent. If the same agent had given a wrong answer, the "Correct and complete answer" auto-fail would make it 0, and the review becomes a coaching conversation about accuracy rather than tone.
How to roll out your scorecard
- Share the scorecard with agents before the first review, so nobody is scored against rules they have not seen.
- Review 4 to 8 random interactions per agent per month, across channels and contact reasons, plus every escalation.
- Calibrate monthly: two reviewers score the same 3 interactions and agree on any line where they differ by more than one step.
- Discuss each review in a short 1:1 with one thing to keep and one thing to change, not a list of every miss.
- Track the team average and each criterion's pass rate: the weakest line across the team is your next training topic.
Limits of a QA score
A QA score measures how an interaction was handled against your own rules, not whether the customer was happy. Read it next to CSAT and first contact resolution: a team with high QA and falling CSAT is usually scoring the wrong behaviours. Small samples swing a lot, so judge an agent on several weeks of reviews, never on one interaction.
Frequently asked questions
- What is a QA scorecard in customer service?
- A QA scorecard is the checklist a team lead or QA analyst uses to review a call, chat or email. Each line is one behaviour (greeting, correct answer, empathy, documentation), each has points, and the total gives a quality score out of 100 that you can compare across agents and weeks.
- How do you calculate a QA score?
- Add the points the agent earned, divide by the points that applied to that interaction, and multiply by 100. Partly met earns half the points and N/A lines are left out of both totals. If the agent fails a critical (auto-fail) item, such as giving a wrong answer or skipping identity checks, the score is 0 whatever the rest.
- What is a good QA score for a call center?
- Most support teams set the target between 80 and 90. Below 70 usually means coaching is needed; 90 and above is excellent. The number only means something if reviewers score the same way, so calibrate: have two people score the same interactions every month and discuss the gaps.
- How many criteria should a QA scorecard have?
- Between 8 and 15. Fewer and you miss behaviours that matter; more and reviews take too long and reviewers start guessing. Put most points on resolution and accuracy, and keep auto-fail items for the few mistakes that hurt customers or break rules.
- How many interactions should I review per agent?
- A common starting point is 4 to 8 interactions per agent per month, picked at random across channels and contact reasons, plus every escalation or complaint. That is enough to see patterns without turning QA into a full-time job for the team lead.
Related
About the author
Abdessamad Ghanem
Customer Experience & Customer Success Consultant · Founder of Clarivoxx
Abdessamad Ghanem works across Customer Support, Sales, Customer Success, Account Management and Partner Management. He writes Clarivoxx Insights for B2B SaaS professionals who own retention, adoption and customer outcomes.