Call center quality assurance (QA) is the process of reviewing customer calls, chats and emails against a written standard, scoring them, and using the results to coach agents and fix processes. A good QA program rests on four things: a clear scorecard, a fair sample of interactions, regular calibrated reviews, and coaching that follows every score. This guide covers each, plus eight best practices, the metrics to track and where AI fits.
If you would rather hand the reviewing to a specialist team, our call center quality assurance service runs this program for consumer brands across voice, chat and email.
Key takeaways for 2026
- Call center quality assurance keeps customer interactions consistent, accurate and on brand.
- A written scorecard, a fair sample and calibrated reviewers matter more than the tool you use.
- Every score should lead to coaching; QA that only grades agents does not improve anything.
- AI can now review far more interactions than a human team, but people still decide what “good” means and do the coaching.
- For consumer brands, QA is how you protect first contact resolution, CSAT and repeat purchase at scale.
What is call center quality assurance and why is it important?
Call center quality assurance evaluates customer interactions to check that they meet a set standard for service quality, compliance, communication and resolution. A modern QA program does more than monitor agents. It measures how well each interaction serves the customer, and it surfaces the patterns behind the scores: recurring customer issues, training gaps and process problems that would otherwise stay hidden in the queue.
It matters because the phone is where customers go when something is urgent. In Forethought’s 2025 survey of 2,243 US adults, 65% said voice is the fastest way to resolve a customer service issue, and 66% said they are most likely to show emotion during a phone interaction. Those are the calls where one bad answer costs a customer.
It also matters because quality and repeat contacts are linked. SQM Group’s research finds that customer satisfaction drops by 15% on average every time a customer has to contact a company again about the same issue. QA is how you catch the wrong answers and half-fixes that cause those second calls.
QA is not the same as quality control. Quality control checks a finished call. Quality assurance builds the standard, the training and the process so that the next call is right the first time.
How a QA program works
A QA program runs in a loop: write the standard as a scorecard, choose which interactions to review, score them, calibrate the reviewers so a score means the same thing whoever gives it, coach each agent on the result, and send the patterns to the teams that own the policy, product or help center. The sections below cover the scorecard and the sample in detail, then the eight practices that keep the loop working.
Building a call center QA scorecard
The scorecard is the heart of the program. Keep it short enough that reviewers score consistently, and specific enough that an agent knows exactly what to change. Most teams use five or six sections with weights that reflect what matters most to the business. Here is a starting template for a consumer brand:
| Section | What the reviewer checks | Example weight |
|---|---|---|
| Opening and verification | Greeting, name, identity and order checks done correctly | 10% |
| Understanding the issue | Active listening, the right questions, no need for the customer to repeat themselves | 15% |
| Accuracy and resolution | Correct answer, policy applied correctly, issue actually solved or properly escalated | 35% |
| Communication and empathy | Clear, warm, on-brand tone; acknowledges frustration; no jargon | 20% |
| Process and notes | Right tags, complete notes, correct follow-up set | 10% |
| Closing | Confirms the fix, explains next steps, offers further help | 10% |
Two rules make the scorecard fair:
- Critical errors are pass or fail. Anything that exposes the customer or the company, such as skipping identity checks, reading back a full card number, or promising a refund policy that does not exist, fails the interaction whatever else went well.
- Each line has a written definition. “Showed empathy” means nothing on its own. “Acknowledged the customer’s problem in their words before moving to the fix” can be scored the same way by any reviewer.
Use the same scorecard across voice, chat and email, with small channel tweaks, so that quality can be compared across the operation.
How many calls to review: the sampling method
Most QA teams cannot listen to every call, so the sample decides what the program sees. A good sample has three layers.
1. A random sample per agent. Pick a fixed number of interactions per agent per week at random, for example five, spread across days and times. This gives every agent a fair, comparable score and stops reviewers choosing calls they already expect to be good or bad.
2. A targeted sample. Add interactions picked on purpose because they carry more risk or more to learn from:
- Every interaction with a low CSAT or a complaint
- Escalations, refunds above a set value, and cancellations
- Calls much longer than the average handle time
- All interactions from agents in their first 30 days
- Repeat contacts about the same order within seven days
3. A statistical check across the operation. To report an operation-wide QA score with confidence, the sample has to be big enough. The standard sample-size formula is n = z² × p(1 − p) ÷ e². At 95% confidence (z = 1.96), with p = 0.5 and a margin of error of 5% (e = 0.05), that works out to about 385 reviewed interactions per reporting period. A larger margin of error needs fewer; a smaller one needs more.
Keep the random and targeted scores separate in reporting. Targeted reviews are chosen because something went wrong, so mixing them in would make average quality look worse than it is.
1. Define clear quality standards
If you do not define what good looks like, agents are left guessing, and inconsistent service follows. The scorecard above is the core of that definition. Share it with every agent, walk new hires through it in training, and show example calls that score high and low on each line. When expectations are clear, performance rises.
Free guide
Customer Service Standards and QA Scorecard: A Playbook for Support Teams
Service standards you can adopt, a QA scorecard and a calibration routine.
2. Monitor and review calls regularly
Review on a steady weekly rhythm, not in a rush at month end. A QA analyst who reviews each agent’s five random calls a week will see patterns early: one agent who nails the greeting but misses the resolution, another who is accurate but sounds rushed. Add speech and text analytics on top to find the trends no human reviewer would spot in a sample, such as a spike in calls about the same delivery issue.
3. Provide constructive feedback
Coaching should feel like support, not a telling off. Good feedback sessions:
- Start by asking the agent how they think the call went
- Name what went well before what to change
- Use the recording or transcript as the example
- End with one specific, practical thing to try on the next shift
Feedback should reach the agent within a few days of the call, while they still remember it.
4. Train agents on soft skills and empathy
Customers want to be heard as well as helped. A technically correct but cold agent can still lose a customer. Train active listening, how to acknowledge frustration, how to explain a “no” clearly, and how to calm an escalating call. Role play the hardest real calls from your QA reviews, since those are the situations agents actually face.
5. Use AI and automation for smarter QA
Manual review of a small sample no longer has to be the whole picture. AI tools can now transcribe and score a much larger share of interactions, flag sentiment shifts, long silences and escalation risk, and route the interactions that need a human reviewer to the top of the queue.
Use AI to widen coverage and to point reviewers at the right calls. Keep people in charge of the scorecard, of calibration and of coaching. An automated score that nobody checks against human judgment can drift quietly away from what your customers actually value.
6. Encourage agent self-assessment and peer feedback
Agents who score their own calls against the scorecard learn the standard faster and accept feedback more readily. Add peer reviews, where agents listen to each other’s best calls, to spread good habits. Make it routine so feedback feels like growth, not scrutiny.
7. Track and improve on key QA metrics
QA is about data, not impressions. The metrics that show whether the program is working:
- QA score: the average scorecard result, by agent, team and channel
- First contact resolution (FCR): the share of issues solved without a follow-up contact. SQM Group puts the call center industry average at 70%.
- Average handle time (AHT): read it next to FCR, since pushing speed alone usually pushes resolution down
- Customer satisfaction (CSAT): the customer’s own view of the interaction
- Repeat contact rate: the share of customers who come back about the same issue within a set window
- Calibration variance: how far reviewers’ scores differ on the same interaction
For a full list of definitions and formulas, see our guide to customer service metrics. Use the results to take action, such as adding staff at peak hours, rewriting a confusing policy or changing call routing.
8. Reward high-performing agents
Recognition drives engagement. QA should feel like a way to be seen doing good work, not only a way to be caught out. Reward consistent top scores, share great calls in team meetings, and use friendly team challenges. Agents who are proud of their QA scores create better experiences, and better experiences keep customers coming back.
Outsourcing call center QA
Running QA in house takes trained analysts, a calibrated scorecard and time that most growing support teams do not have. Many brands hand it to a specialist partner, either to review their in-house team or as part of a fully outsourced support program.
At Atidiv, our average QA score is 98%, with a 4.8 average CSAT rating. Dedicated QA analysts review tickets and calls against scorecards customized to each brand’s standards, with real-time feedback, weekly coaching sessions and monthly calibration with the brand’s team. We also work to SOC 1, GDPR, CCPA and ISO standards. See how the reviews and reporting work on our quality assurance services page, or explore our call center outsourcing teams.
Free guide
Customer Service Standards and QA Scorecard: A Playbook for Support Teams
Service standards you can adopt, a QA scorecard and a calibration routine.
Frequently asked questions on call center quality assurance
How do you measure call quality?
Call quality is measured by scoring a sample of calls against a written scorecard that covers accuracy, resolution, communication and process, then reading that QA score alongside first contact resolution, average handle time and CSAT. The scorecard tells you what the agent did; the other metrics tell you what it achieved.
How many calls should a QA team review per agent?
Around five random interactions per agent per week, plus targeted reviews of complaints, escalations and new hires; the sampling method above shows how to size the operation-wide sample.
What should a call center QA scorecard include?
Six weighted sections, with accuracy and resolution weighted most and compliance errors as automatic fails; the template above gives a starting set of weights.
What’s the best way to give agents feedback without making them feel micromanaged?
Make it a coaching conversation. Start with what went well, use the actual recording, and agree on one specific change. Letting agents score their own calls first makes feedback land far better than a manager’s verdict alone.
Can AI replace human call center quality assurance?
No. AI is good at transcribing, spotting patterns and scoring large volumes of interactions. People are still needed to set the standard, calibrate scores, judge difficult calls and coach agents. The strongest programs use both.
Why should businesses outsource their call center QA?
Outsourcing gives you trained analysts, a proven scorecard and consistent weekly reviews without building a QA team in house. It also brings an outside view, which helps when internal teams are too close to their own processes.
Ashish is Atidiv's President, and his mission is to help consumer brands leverage the post-purchase customer experience as a growth opportunity. With 15 years of strategy and operations experience, he uses analytics and a hands-on approach in building strong teams, developing scalable and effective systems, and tackling challenging business problems with practical solutions.