How to size your quality monitoring team: the formula

How many QA analysts your operation needs: the formula using interaction volume, AHT and productive hours, three worked examples and common mistakes.

BRIGGS Team6 min readLer em português

Introduction

Ask a call center manager how they arrived at the number of QA analysts on their team. Most of the time the answer is historical: it started with two, grew to four when the operation doubled, and stayed there.

There is a formula. It has three variables and fits on one line. This article shows the formula, works through three real scenarios and points out the mistakes that throw sizing off.

The formula

The number of QA analysts you need comes from:

QA analysts = (monthly volume × desired coverage) ÷ audits per analyst

The first term is how many audits you need to do. The second is how many each person can deliver. That's it.

The hard part is getting the second number right, and that's where almost every sizing exercise goes wrong.

How much a QA analyst really audits

The temptation is to divide working hours by AHT (average handle time). If the analyst works 132 hours and a call lasts 6 minutes, that would be 1,320 audits a month. That number never happens.

Three deductions need to go in.

Productive hours. Of the 132 monthly hours, about 92 are actually spent auditing. The other 40 go to calibration, agent feedback, quality meetings and reporting. That time is part of the job, not slack.

Time per audit. Auditing a call takes about 2.5 times its length. You have to listen, often go back over parts of it, fill out the form criterion by criterion, write a justification and log it. With a 6-minute AHT, each audit takes 15.

Availability. Vacation, sick leave and absences cut team capacity by about 10% over the year. A team of four runs, on average, with 3.6 people.

The real math:

92 hours × 60 = 5,520 productive minutes
6 min AHT × 2.5 = 15 minutes per audit
5,520 ÷ 15 = 368 audits per analyst/month (gross capacity)
368 × 90% availability = 331 effective audits

Keep both numbers. Gross capacity is for setting individual targets; effective capacity is for sizing the team. Between 300 and 400 audits a month per analyst is the realistic range for an operation with an AHT between 5 and 8 minutes.

Three worked scenarios

Small operation: 5,000 interactions/month

Coverage target: 10%, which means 500 audits.

500 ÷ 331 = 1.5 QA analysts

Two analysts cover it comfortably and still have capacity left for cross-calibration, which is what keeps the yardstick consistent. At this size, the manual model works well.

Mid-size operation: 30,000 interactions/month

With four QA analysts, after vacation and absenteeism (3.6 available on average over the year), capacity is 1,325 audits, or 4.4% coverage.

To reach 30% coverage:

30,000 × 0.30 = 9,000 audits
9,000 ÷ 331 = 27.2 QA analysts

Twenty-seven analysts would cost about $209,000 a month, at the $93,000 a year that a US QA analyst costs the employer once benefits are included (the BLS-based figure our calculator uses). And 70% of the operation would still go unaudited.

Large operation: 100,000 interactions/month

For 99% coverage:

100,000 × 0.99 = 99,000 audits
99,000 ÷ 331 = 299 QA analysts

Two hundred ninety-nine QA analysts, more than $2.3 million a month in quality payroll alone. No operation keeps a team that size, which is why high coverage simply doesn't exist in the manual model.

You can run these numbers with your own operation's data in the quality monitoring coverage calculator.

The five mistakes that break the sizing

Using full working hours as the base. Sizing on 132 hours instead of 92 inflates capacity by 43%. The team starts the month with a target it can't hit, and the backlog of pending audits grows.

Ignoring the 2.5 factor. Anyone who assumes audit time equals AHT underestimates it by a factor of more than two. It's the most common mistake and the most expensive one.

Forgetting vacation and absenteeism. A team of four people operates, in practice, with 3.6 over the year. If you size for the perfect scenario, you're short-staffed one month in three.

Treating all channels the same. Auditing a WhatsApp chat doesn't cost the same as auditing a 12-minute call. If the operation is multichannel, size each channel separately and add them up.

Not separating auditing from coaching. If the same person audits and runs feedback sessions, coaching time has to come out of productive hours. Teams that don't make this separation find out mid-quarter that coverage dropped and nobody knows why.

What coverage to aim for

There is no Brazilian standard that sets a minimum sampling percentage. What exists are market practices and, in regulated industries, traceability requirements that a sample can't meet.

A useful yardstick: ask what share of problem calls you're willing to never find out about. With 5% coverage, the answer is 95%.

In a regulated operation, this stops being a quality question and becomes an exposure question. A NIP from the ANS (a formal complaint notice from Brazil's health plan regulator) arrives pointing to a specific call. If it was in the 95% that wasn't audited, your defense starts with no evidence.

When the math stops making sense

Manual sizing works up to a certain volume. The sign that you've passed that point shows up in the math before it shows up in the budget.

When the number of QA analysts needed for your desired coverage approaches the size of your customer service team, the discussion has changed in kind. The question is no longer headcount; it's the limit of the model.

One operation running BRIGGS solved this in the opposite direction: it went from eleven QA analysts to four and raised coverage from 5% to 99%. The seven who were reassigned moved on to analyzing trends, calibrating criteria and running coaching, work that the listening routine had kept them from doing.

Conclusion

Sizing a quality monitoring team isn't an exercise in intuition. There are three variables: volume, desired coverage and real capacity per QA analyst.

Real capacity falls between 300 and 400 audits a month per person, and anyone using a higher number is sizing for a scenario that doesn't happen.

Do the math for your operation before the next headcount discussion. If the result points to a quality team bigger than your customer service team, the problem isn't the number of people. It's the model.

Calculate your operation's coverage and see how many QA analysts it would take to close the gap.

Related reading