Publication/Call Center QA Scorecard: How to Build One That Improves CSAT

Call Center QA Scorecard: How to Build One That Improves CSAT

Faizan Kanth
AuthorFaizan Kanth
PublishedJune 11, 2026
Read Time9 min read
Call Center QA Scorecard: How to Build One That Improves CSAT
call center QA scorecardquality assurancecall quality monitoringCSAT improvementagent coaching

Quick answer: A call center QA scorecard is a short form that scores each customer interaction against the behaviors and outcomes you care about. Build it around four to six weighted categories, add clear auto-fail rules for compliance, calibrate your reviewers regularly, and feed every score into coaching. If scores do not move in step with CSAT and first contact resolution, the scorecard is measuring the wrong things.

Most support teams already have a scorecard. Far fewer have one that changes how agents work. The usual version has grown by committee over several years, rewards reading a script correctly, and produces a number that everyone reports and nobody acts on.

This guide shows how to build a scorecard from scratch or rebuild the one you have, with a sample template you can adapt.

What is a call center QA scorecard?

A QA scorecard, sometimes called a call quality monitoring form, is the tool reviewers use to evaluate a call, chat or email. It lists what a good interaction looks like, assigns points to each element, and produces a score for the interaction and, over time, for the agent and the team.

A good scorecard does three things. It tells agents exactly what is expected. It gives team leaders specific evidence for coaching. And it gives managers a quality measure that predicts what customers will say in surveys.

The third point is the test. If your QA scores are high and CSAT is flat or falling, the form is scoring behaviors customers do not care about.

Why do most QA scorecards fail?

  • They are too long. A form with dozens of line items takes too long to complete, so fewer interactions get reviewed and reviewers start skimming.
  • They score the script, not the result. Using the customer's name three times earns points. Fixing the problem earns the same number, or fewer.
  • Every item is yes or no. Binary scoring cannot separate an adequate answer from an excellent one.
  • Nobody calibrates. Two reviewers score the same call differently, agents notice, and trust in the program disappears.
  • The sample is too small to be fair. One bad call in a tiny sample can sink an agent's month.
  • Nothing happens next. Scores go into a report. Coaching, if it happens at all, is not tied to what the scores showed.

How to build a call center QA scorecard in seven steps

  1. Start from outcomes. Write down the two or three results the scorecard should drive, such as first contact resolution, CSAT and compliance. Every item on the form should connect to one of them.
  2. Choose four to six categories. Typical categories are opening and verification, discovery, resolution accuracy, compliance, communication and empathy, and wrap-up. Fewer categories make the form faster to use and easier to coach from.
  3. Write observable criteria. "Showed empathy" is an opinion. "Acknowledged the customer's problem before asking for account details" can be heard on a recording. Aim for two to four criteria per category.
  4. Weight the categories. Give the most points to what matters most to the customer, which is nearly always whether the issue was resolved correctly. Weights should add up to 100.
  5. Define auto-fails. A small set of errors should fail the interaction regardless of the rest. Keep these to compliance, security and conduct.
  6. Pick a rating scale. A three-point scale (met, partly met, not met) is usually enough. It captures nuance without inviting endless debate between a 6 and a 7.
  7. Pilot and calibrate. Have several reviewers score the same ten interactions, compare results, and rewrite any criterion they interpreted differently before you launch.

A sample QA scorecard template

The template below is a starting point for a voice support team. Adjust the criteria and weights to your product and your customers.

CategoryExample criteriaWeight
Opening and verificationClear greeting, identity verified per policy10
DiscoveryAsked the right questions, confirmed the issue15
Resolution accuracyCorrect fix, right process, no repeat contact needed30
Compliance and securityRequired disclosures made, data handled correctly15
Communication and empathyClear language, acknowledged the customer, right pace20
Wrap-up and documentationNext steps confirmed, accurate ticket notes10

A few notes on using it:

  • Scoring. Rate each criterion as met, partly met or not met, and award full, half or zero points for that line.
  • Auto-fails. Typical examples are skipping identity verification, disclosing one customer's data to another, omitting a legally required statement, and rude or abusive conduct. An auto-fail sets the interaction score to zero and triggers an immediate review.
  • Pass mark. Set it from your own baseline. Score a month of interactions with the new form, see where your strongest agents land, and place the threshold just below that level.
  • Comments. Require one sentence of evidence for any line not fully met. It makes coaching faster and disputes rarer.

Resolution accuracy carries the most weight on purpose. A warm, well-structured call that gives the wrong answer is a failed call.

How many interactions should you score, and how do you calibrate?

Score enough interactions per agent that one unusual call cannot define the result. A fixed number per agent each week, selected at random, is fairer than a monthly batch chosen by the team leader. Add a targeted sample on top: low CSAT responses, long handle times, repeat contacts and escalations. Random samples show you the typical interaction. Targeted samples show you where things break.

Calibration keeps scoring consistent. Bring reviewers, team leaders and, if you outsource, your vendor's quality analysts together on a regular schedule. Everyone scores the same interactions independently, then the group compares results line by line. Where scores differ, agree on the correct reading and update the written guidance.

Track the gap between reviewers over time. A shrinking gap means the criteria are clear. A persistent gap usually points to a criterion that is still an opinion in disguise.

How do you adapt the scorecard for chat and email?

Keep the same categories and change the criteria. Customers judge written channels differently, and the form should reflect that.

  • Chat. Add criteria for response pace within the conversation, handling of concurrent chats without long silences, and correct use of saved replies. A saved reply that does not match the question should lose points.
  • Email. Score completeness, because a good email answers every question the customer asked and anticipates the next one. Add criteria for structure, tone and accuracy of any links or attachments.
  • Both. Grammar and spelling matter, but weight them lightly. A flawless reply that misses the point still fails.

Use one scorecard family across channels so that scores stay comparable and agents who work several channels are not learning three different standards.

How do you turn QA scores into better CSAT?

Scoring is measurement. Improvement comes from what happens afterward.

Tie every review to a coaching conversation within a few days, while the agent still remembers the interaction. Focus each session on one or two behaviors, play the recording, and agree on what the agent will do differently. Then check the next set of scores for that specific line. If it did not move, the coaching did not work, and the team leader needs a different approach.

At team level, look for patterns. If discovery scores are low across the board, the problem is training or the knowledge base, not individuals. If one category scores well and CSAT is still weak, revisit whether that category measures something customers value.

Finally, compare QA scores against CSAT and first contact resolution for the same interactions. They should rise and fall together. This is far easier when conversation history, survey results and quality data sit in one place, a point we cover in our post on how a customer engagement platform turns every interaction into a growth opportunity.

Where does AI-assisted QA fit?

Manual QA reviews a sample. AI-assisted QA can transcribe and score every interaction against the same criteria, which changes what a quality team is able to see.

With full coverage, you stop guessing whether a sampled call was typical. You can spot a compliance miss on the day it happens, track a single criterion across thousands of contacts, and direct human reviewers to the interactions that most need attention.

AI scoring has limits. It is reliable on objective items, such as whether a disclosure was read, and weaker on judgment calls such as whether an agent was right to bend a policy. The sensible model is AI scoring everything, with people reviewing exceptions, auto-fails and disputed scores, and people leading every coaching conversation.

If you are evaluating an outsourcing partner, ask how each vendor combines the two. Our list of customer support outsourcing RFP questions includes the quality questions worth putting in writing.

How SSGSERV runs quality assurance

Quality is a managed process in our centers, not a monthly report. Our customer experience teams follow COPC-aligned practice, and our operations hold ISO 9001, ISO 27001 and CMMI Level 3.

  • Scorecards built with you. We design or adopt the scorecard for each program with the client, weight it toward resolution, and calibrate jointly so your team and ours score the same way.
  • Dedicated monitoring. Our call center quality monitoring service covers systematic evaluation, feedback and coaching, for our own delivery teams or as a stand-alone review of yours.
  • AI with people in charge. Agent K, our AI knowledge and quality assurance system, is designed to give supervisors structured insight while they keep authority over quality decisions. It is currently in limited early access.

You can see the full scope of our call center quality assurance programs on the service page.

Frequently asked questions

What should be on a call center QA scorecard?

Four to six weighted categories that cover the interaction from start to finish: opening and verification, discovery, resolution accuracy, compliance, communication and wrap-up. Each category needs two to four observable criteria. Add a short list of auto-fail errors for compliance and security.

What is a good QA score in a call center?

There is no universal number, because every scorecard is weighted differently. Set your pass mark from your own baseline, using the level your strongest agents reach consistently. A good score is one that rises alongside CSAT and first contact resolution.

How many calls should QA review per agent?

Enough that a single unusual call cannot decide the result. A fixed number of randomly selected interactions per agent each week works better than a larger monthly batch. Add targeted reviews of low CSAT contacts, repeat contacts and escalations.

What is an auto-fail in call center QA?

An auto-fail is an error serious enough to fail the whole interaction regardless of other scores. Common examples are skipping identity verification, mishandling customer data and missing a required disclosure. Keep the list short so that it stays meaningful.

How is a QA scorecard different from CSAT?

CSAT records how the customer felt. A QA scorecard records what the agent did, including things the customer cannot see, such as compliance and documentation. You need both, and they should move in the same direction.

To see how a governed quality program fits together, from scorecard design to calibration and reporting, download our CX Quality Governance Blueprint. If you would like a second opinion on your current scorecard, get in touch with our team and we will review it with you.

Article By
Faizan Kanth

Faizan Kanth founded SSGSERV with a vision to build a scalable global delivery model, connecting Bangalore, one of India's leading technology hubs, with Srinagar, an emerging center for skilled tech talent. Under his leadership, SSGSERV has grown into a specialized operations partner for global technology companies, serving clients across the United States, the Middle East, the United Kingdom, and the Philippines. With over a decade of experience across leading global technology, automotive, and business process organizations, Faizan brings deep expertise in operations management, business strategy, and market evaluation to every engagement. A postgraduate of a premier Indian management institute, he is driven by an entrepreneurial approach to solving complex operational challenges for enterprises worldwide. Beyond building SSGSERV, Faizan is focused on positioning Kashmir as a hub for Global Capability Centers, having grown the company's Srinagar delivery center to over 500 professionals and counting.

Share this story
Back to all insights

Trust & Assurance

Technology Partners

Gartner
Partner
Dell
Partner
Cisco
Partner
HPE
Partner
Clutch
Partner

Governance & Compliance

ISO Certified Operations
[ STATUS: COMPLIANT ]
[ NETWORK: SOVEREIGN ]
[ GOVERNANCE: ENTERPRISE ]