Customer Support QA Scorecard: A Template and How to Use It

Response times tell you how fast your team answers. They say nothing about whether the answer was right. A quality assurance (QA) scorecard closes that gap: a short, fixed checklist a reviewer uses to grade real conversations, so quality becomes a number you can track, coach and improve.

TL;DR

• A QA scorecard grades a sample of support conversations against the same weighted criteria every time.
• Use four categories: Accuracy (30%), Resolution (30%), Process (20%) and Tone (20%).
• Add auto-fail rules for the mistakes that cost money or trust, such as a wrong refund promise or sharing data with the wrong person.
• Score = points earned ÷ points possible × 100, leaving out criteria that do not apply.
• Review at least 5 conversations per agent per week, aim for a team score of 85% or higher, and calibrate reviewers monthly.

What is a customer support QA scorecard?

A QA scorecard is a rubric for reviewing support conversations. A team lead or QA reviewer picks a sample of tickets, chats and WhatsApp threads, checks each one against a list of criteria and records a score. Because every reviewer uses the same list, scores are comparable across agents, weeks and channels.

It measures what customers experience but rarely report: whether the answer was correct, whether the problem was actually solved, and whether the agent followed the process that protects the business.

The QA scorecard template

Score each criterion Yes (full points), Partly (half points), No (zero) or N/A when it does not apply to the conversation.

1. Accuracy (30 points)

• The answer was factually correct and matched current policy and pricing. (15)
• The agent answered every question the customer asked, not just the first one. (10)
• Links, attachments and instructions were correct and worked. (5)

2. Resolution (30 points)

• The customer's problem was solved, or the next step and its timing were clearly stated. (15)
• The agent resolved it without unnecessary back-and-forth or transfers. (10)
• The ticket was closed only when the issue was actually resolved. (5)

3. Process (20 points)

• Priority, category and tags were set correctly. (5)
• Internal notes record what was done and why, so a teammate can pick it up. (5)
• The agent checked the customer's history before replying. (5)
• Escalation and handoff followed the team's rules. (5)

4. Tone (20 points)

• Clear, plain language with no jargon the customer would not know. (5)
• Polite and empathetic, matched to the customer's mood. (5)
• Correct spelling, grammar and the customer's name. (5)
• The reply was easy to scan: short paragraphs, steps numbered. (5)

Auto-fail (score the whole conversation 0)

• Promised a refund, discount or delivery date the agent was not authorised to give.
• Shared personal or account data with someone who had not been verified.
• Rude, dismissive or unprofessional language.

Keep auto-fail rules few and serious. If everything is a critical error, nothing is.

How do you calculate a QA score?

Add the points earned and divide by the points possible for that conversation, leaving N/A criteria out of both:

QA score = points earned ÷ points possible × 100

Example: a WhatsApp chat where escalation did not apply has 95 points possible. The agent earned 81, so the score is 81 ÷ 95 × 100 = 85.3%.

The team score is the average across every reviewed conversation in the period. Report it beside the share of conversations that scored 0 on an auto-fail, because one auto-fail matters more than a few points of tone.

How many conversations should you review?

Review at least 5 conversations per agent per week, or 2 to 5% of total volume, whichever is higher. Pick them at random within each channel so you see email, chat and WhatsApp work, and add every conversation that received a Poor rating. Those are where the coaching value is.

New agents need more reviews in their first month. Experienced agents scoring consistently above 90% can drop to a lighter sample.

What QA score targets are reasonable?

• Team QA score: 85% or higher.
• Individual agents: 80% minimum, with coaching for anyone below it two months running.
• Auto-fails: zero is the target; investigate every one.

These match the quality targets in our SLA examples for customer support, which set the response and resolution targets this scorecard sits beside. Speed and quality belong in the same monthly review.

How do you keep reviewers consistent?

Run a calibration session once a month. Every reviewer scores the same three conversations on their own, then the group compares results criterion by criterion. Where scores differ by more than 10 points, agree what "Yes" means and write a one-line example into the scorecard. After two or three sessions, reviewers score within a few points of each other.

How do you turn QA scores into better support?

• Share the scorecard with agents first. Nobody should be graded on criteria they have never seen.
• Coach on the lowest category, not the lowest score. An agent at 78% overall may be strong on tone and weak on accuracy. Train the gap.
• Fix the system when the whole team misses. If most agents fail the same accuracy criterion, the knowledge base article or the policy is unclear, not the agents.
• Compare QA with CSAT. Conversations that score high on QA but receive Poor ratings point to expectations your process does not meet.

Where do QuiQ Help Desk teams find the data?

QuiQ Help Desk keeps everything a reviewer needs on one ticket: the full conversation on every channel, internal notes, priority and category, SLA deadlines, and the customer's Happy, Fair or Poor rating. The customer panel shows their past tickets, open deals and ratings, so reviewers can check whether the agent used the context available. Pull your weekly sample from the queue, include every ticket rated Poor, and score them against the template above.

For measuring satisfaction on chat channels specifically, read how to measure customer satisfaction (CSAT) on WhatsApp.