Building a Quality Assurance Program

A quality assurance (QA) program is how a contact center makes sure customers get accurate, consistent, helpful service, every time and from every agent. Done well, it’s one of the most valuable tools you have. Done poorly, it becomes a box-checking exercise that frustrates agents and changes nothing.

This guide walks through building a program that actually improves results.

What a quality program should do

A strong QA program serves four purposes:

1. Protect the customer experience by making sure contacts are handled accurately and professionally
2. Develop agents through specific, regular coaching
3. Find process problems, such as confusing policies or broken systems, that cause poor experiences
4. Reduce risk by checking that legal and compliance requirements are followed

If your program does only the first and fourth, it will feel like policing. The second and third are what make it worthwhile.

Step 1: Define what quality means for you

Before building a scorecard, agree on what a great customer experience looks like at your company. Bring together leaders, supervisors, trainers, and a few experienced agents, and ask:

• What do our customers value most: speed, accuracy, friendliness, getting it fixed the first time?
• What must happen on every contact, such as verifying identity or giving required disclosures?
• What do our best agents do that others don’t?
• What do customers complain about most?

Look at customer feedback, too. Survey comments and complaints show what customers notice, which isn’t always what internal teams expect.

Step 2: Design a simple scorecard

Your scorecard turns that definition into specific, observable behaviors. Keep it short, usually five or six sections with 10 to 15 items total.

A common structure:

Section	Example items
Opening	Greeting, identity verification
Understanding	Asked clarifying questions, confirmed the issue
Resolution	Accurate information, solved the issue or set clear next steps
Communication	Clear language, empathy, professional tone
Process and compliance	Required steps, disclosures, accurate notes
Closing	Summarized next steps, confirmed nothing else needed

Design tips:

• Describe each item clearly, with examples of what “meets” and “doesn’t meet” look like. Vague items like “good attitude” produce inconsistent scores.
• Weight items by impact. Accurate resolution should count more than using the customer’s name.
• Separate critical items, such as compliance failures, so a high overall score can’t hide them.
• Use simple ratings, like Yes / No / Not applicable, which are easier to apply consistently.
• Create separate versions for calls, chats, and emails if the channels work differently.

Our guide to running call reviews shows how supervisors use the scorecard day to day.

Step 3: Decide who reviews and how often

Who reviews:

• Supervisors review their own team’s contacts, which connects quality directly to coaching.
• Dedicated QA specialists, common in larger centers, add consistency and free up supervisor time.
• Agents review their own contacts as a self-assessment.

Many centers use a mix: QA specialists for consistent scoring, and supervisors for coaching.

How many contacts to review: start with a set number per agent each month, the same for everyone. A handful of thoroughly reviewed contacts per agent is more useful than many rushed ones. Increase reviews for new agents and anyone working on a specific skill.

Which contacts: mix random contacts, which show typical performance, with targeted ones, such as long contacts, escalations, complaints, and low survey scores.

Step 4: Build in calibration

Calibration means having several reviewers score the same contact independently, then comparing and discussing their scores. It keeps scoring consistent and fair.

• Hold calibration sessions regularly, often monthly.
• Track how far apart reviewers’ scores are, and aim to narrow the gap over time.
• Update scorecard descriptions whenever confusion comes up.
• Invite experienced agents occasionally, which builds trust in the program.

Without calibration, agents quickly notice that scores depend on who reviewed them, and the program loses credibility.

Step 5: Connect results to action

Quality scores are only useful if something happens next:

• Individual coaching: every review should lead to a conversation, starting with strengths and focusing on one area to improve. See Coaching conversations that work.
• Training updates: when many agents struggle with the same skill, update training for everyone, including new hires.
• Process fixes: when agents follow the process correctly but customers still have a poor experience, the process is the problem. Share those findings with the teams who own it. See Process improvement for contact centers.
• Knowledge base updates: when agents give inconsistent answers, the knowledge base may be unclear or outdated.

Step 6: Create a fair dispute process

Agents should be able to question a score they believe is wrong. A simple process might be:

1. The agent explains their concern in writing within a set time, such as five working days.
2. A second reviewer, or the QA lead, listens to the contact and decides.
3. The result, and the reasoning, is shared with the agent.

A fair dispute process builds trust and often reveals unclear scorecard items.

Step 7: Report what matters

Share quality results at three levels:

• Agents: their own scores, trends, and specific feedback
• Supervisors: team results, common gaps, and calibration results
• Leadership: overall quality, compliance issues, top process problems, and actions taken

Compare quality scores with customer measures, such as satisfaction and first contact resolution. If quality scores are high but customers are unhappy, your scorecard may be measuring the wrong things.

Using technology

• Quality management software organizes reviews, scorecards, calibration, and reporting in one place.
• Speech and text analytics can scan every contact for keywords, long silences, negative sentiment, or missing compliance phrases, then flag the contacts most worth reviewing.
• AI-assisted scoring can pre-score some items automatically, letting reviewers focus on judgment and coaching.

Technology can greatly expand how many contacts you learn from. But human reviewers are still needed for tone, context, and fairness, and for the coaching conversation that makes quality improve.

Launching your program

1. Form a small team to define quality and draft the scorecard.
2. Test the scorecard on 20 to 30 real contacts and refine unclear items.
3. Hold calibration sessions before scores count.
4. Explain the program to agents: why it exists, how it works, and how it helps them.
5. Run a trial period where scores are shared but not tied to performance reviews.
6. Launch fully, then review the program itself every six to twelve months.

Common mistakes to avoid

• Measuring script compliance instead of customer outcomes.
• Long, complicated scorecards that reviewers can’t apply consistently.
• Using QA only to discipline agents, rather than to develop them.
• Skipping calibration, which makes scores feel arbitrary.
• Ignoring process problems that show up in reviews.
• Never updating the scorecard as products, policies, and customer needs change.

What to read next

• Running call reviews: how supervisors review contacts day to day
• Coaching conversations that work: turning quality results into improvement
• Process improvement for contact centers: fixing the problems quality reviews uncover

---