Skip to content
Lines open
Map my coverage

Quality Research

The evidence limits of average handle time as a quality measure

Average handle time describes workload and says little about resolution, so a quality review needs a second measure and a stated limit.

Illustration for The evidence limits of average handle time as a quality measure
Table of contents

Published November 4, 2026

Average handle time is a useful capacity measure and a poor quality measure on its own. It tells a supervisor how long calls take and says nothing about whether the caller's problem was solved. A team can lower the average by rushing the close, and the result may look efficient while repeat contacts rise. A quality review needs a second measure and a clear statement of what the average can support.

Question and scope

This review asks what average handle time can establish about the quality of a call and where the measure stops. The unit is one call that reached an agent. The observation window is the call itself plus a short follow-up for repeat contact. The review excludes calls that never reached an agent and calls that were abandoned in the queue.

The scope also excludes the agent's tone and the accuracy of the answer. Those require a separate review with a rubric and a sample. The report focuses on the relationship between handle time and the outcomes the team can observe. The distinction keeps the capacity question separate from the quality question.

What the measure includes

Handle time usually includes talk time, hold time, and after-call work. The exact definition varies by system and by team. Some systems include the time spent transferring a call, and some exclude it. Some include the time an agent spends researching after the caller hangs up, and some record it as a separate activity. The definition should be written down before the number is compared.

A change in the definition can look like a change in performance. If after-call work is added to the measure, the average rises without any change in the conversation. If a transfer is excluded, the average falls without any improvement in the caller's experience. Record the definition beside the result and keep it stable across the comparison.

Why the average hides the mix

The average combines calls of different types and lengths. A queue with many short calls and a few long complex calls can have the same average as a queue with a uniform call length. The two queues need different staffing and different coaching. The average alone cannot tell them apart.

The distribution is more informative than the average. Report the median, the longest quartile, and the share of calls above a threshold. A small number of very long calls can pull the average up and hide a large group of short calls. The distribution shows the shape and points to the calls that need a different process.

A call that runs long because the caller is upset is different from a call that runs long because the agent cannot find the record. The first may need a supervisor, and the second needs a system fix. Read a sample of the long calls and code the reason before changing the process. The code is more useful than the duration alone, and it gives the supervisor a specific coaching topic.

Record the time spent after the caller hangs up as a separate field. A short conversation with a long note may be a documentation problem, and a long conversation with a short note may be a listening problem. The two patterns call for different coaching and different tools. The separate field makes the difference visible without changing the total measure, and it helps the team explain why the average moved.

Review the longest calls with the agent rather than about the agent. Ask what made the call difficult and what would have helped. The answer may be a missing permission, an unclear policy, or a caller who needed a specialist. Those fixes are often cheaper than a new script, and they are more likely to change the next call.

MeasureWhat it describesWhat it can hide
Average handle timeTotal time divided by callsThe mix of short and long calls
Median handle timeTypical call lengthThe tail of very long calls
Longest quartileCalls that take the most timeWhether the time was useful
Repeat contact rateWhether callers returnedThe reason for the return
Resolution rateWhether the issue closedThe accuracy of the answer

Read the average with the outcome

Pair the handle time with a resolution measure and a repeat contact measure. A team with a low average and a high repeat rate may be closing calls too quickly. A team with a high average and a low repeat rate may be spending time on complex issues that need it. The pair is more useful than either number alone.

Do not assume that a longer call is better. A long call can be thorough, and it can also be inefficient. The quality review should sample the calls and read the conversation. The average points to the calls worth reading, and the reading is what tells the team whether the time was well used.

Compare agents fairly

An agent's average is affected by the calls they receive. A new agent may get simpler calls and a lower average, while an experienced agent may get the complex escalations. A comparison that ignores the call mix will reward the agent with the easier queue and penalize the one who handles the hard cases. Segment the comparison by call type before drawing a conclusion about an individual.

Segment the comparison by call type and by queue. If the cells are too small, report the counts and avoid a ranking. A supervisor can use the pattern to decide who needs coaching and which calls need a different route. The goal is a better process, not a league table.

Where the evidence stops

The average cannot show that a caller was satisfied. A short call can be polite and complete, and a long call can end without a resolution. The measure describes time, not outcome. The review should state that limit wherever the number is published.

The average also cannot show that a change in staffing will improve service. Adding agents may reduce the wait and leave the handle time unchanged. Reducing the handle time may increase the repeat contacts and the total work. The measure is one input to a decision, not the decision.

A protocol for one queue

Choose a month and define handle time, the call types, and the outcome measures. Export the call data and reconcile it with the schedule and the queue. Review the distribution and select a sample of short, typical, and long calls. Read the sample with a rubric that covers the answer, the close, and the next step.

Report the average, the median, the longest quartile, the repeat rate, and the resolution rate. Add a short note on the call mix and the definition. Repeat the review after a script or routing change with the same definitions. The comparison is more useful than a single month.

Limitations and next action

The sample is one queue and one month, and the results depend on the call mix and the systems. A small queue may have too few calls to support a stable average, and a count is more honest than a percentage. The protocol should be repeated before it is used as a baseline. A single review is a description, not a benchmark.

The next action is to add one outcome measure before changing the scorecard. A repeat contact count or a resolution sample is enough to show whether a lower average is helping the caller. The team can then test a change and see both numbers. That is a more useful question than whether the average is high.

Sources

Map the first queue

GIVE MISSED CALLS A WAY BACK IN.

Share your call types, hours, tools, and escalation rules. We will turn them into a practical coverage plan with a clear first week.

Book a free consultation ↗See the call lanes