Average handle time is the wrong target
Paul Szerszen, SVP of Professional Services & Customer Operations
Average handle time is the most reported metric in contact centers and one of the most misapplied. It is a genuinely useful number in exactly one place and a corrosive one nearly everywhere else.
The distinction is not whether to track it. Track it. The distinction is what you do with it. Used as a capacity planning input, average handle time is essential. Used as an agent performance target, it reliably produces the behavior you least want: contacts closed before they are resolved, customers who call back, and a service level that looks better than the experience.
What average handle time is for
Average handle time is talk time plus hold time plus after-contact work, averaged across contacts. Its real job is arithmetic.
To staff a contact center you need to know how much work arrives and how long each unit takes. Forecast volume gives you the first. Average handle time gives you the second. Multiply them and you have workload, which converts into required staffing by interval. Without a reliable AHT figure you cannot build a schedule, size a team, or price a program.
That is the use case. It is a planning input describing a population of contacts. The trouble starts when a number designed to describe a population gets applied to an individual.
Why it fails as an agent target
Tell an agent that their handle time should be under six minutes and you have told them what to optimize. They will optimize it, and the available methods are mostly bad.
An agent under handle time pressure can end the contact before the customer's actual question is resolved. They can skip the confirmation step that catches a misunderstanding. They can transfer a contact that they could have handled. They can avoid asking the follow-up question that would have surfaced the second issue. Each of those lowers handle time. Each of them raises the odds the customer contacts you again.
The cost does not disappear. It moves. It shows up as repeat contacts, lower first call resolution, and escalations, usually in a different report reviewed by different people. A contact center can hit its handle time target and increase its total workload at the same time, and the two facts rarely get connected because they live on separate dashboards.
There is a second cost that is harder to see. Handle time pressure changes how agents talk to customers. The consultative question, the pause that lets a confused caller explain properly, the moment of acknowledgment when something has gone wrong for the customer, are all expensive in seconds and are usually what the customer remembers.
The variation is the signal, not the average
Here is where average handle time becomes genuinely useful at the agent level, and it is not the average.
Look at the distribution rather than the mean. An agent whose handle time is consistent across contact types, and roughly in line with peers on the same contact mix, is probably working the way you would want. An agent whose handle time is far above peers on one specific contact driver has a knowledge gap on that topic, which is a coaching conversation with a clear subject. An agent whose handle time is far below peers with a lower first call resolution rate is closing contacts early.
That is three different conversations, and none of them is "your AHT is too high." The average alone cannot distinguish between them, which is why managing to it produces generic pressure rather than useful coaching.
Contact mix matters here too. Handle time varies enormously by contact driver. An agent routed more complex work will show higher handle time for reasons that have nothing to do with their performance. Comparing raw averages across agents handling different mixes is not a fair comparison and agents know it, which is part of why the metric erodes trust.
What to measure agents on instead
Replace the handle time target with a small set of outcome measures, and keep handle time as diagnostic context.
First call resolution is the primary one. It is harder to game than handle time, because closing a contact early to save seconds shows up directly as a repeat contact. Our guidance on reaching 90 percent FCR and above covers how to build toward it. Quality score against a scorecard that measures whether the customer's issue was actually addressed. Customer satisfaction where you collect it reliably. And adherence to schedule, which is an availability measure rather than a speed measure and is a legitimate expectation.
Mpathic's published work on a statewide education benefits program reached first call resolution of 93 percent against a contractual standard of 80 percent, with customer satisfaction at 4.88 out of 5.0 and average speed to answer of 12 seconds (State agency contact center case study). Those are the measures worth committing to contractually. Handle time is how you staff to meet them.
Where handle time should be reduced, and how
None of this means long contacts are good. It means handle time should be reduced by removing work, not by hurrying agents.
The reducible parts are usually structural. After-contact work that could be automated or eliminated. Hold time spent looking up information that should have been on screen. Time lost navigating between systems that do not talk to each other. Repeated explanation caused by a transfer that lost context. Knowledge that exists but cannot be found quickly.
Each of those is a process fix that lowers handle time without asking an agent to rush anyone. ISO 18295-1:2017, the international standard for customer contact centers, frames performance in terms of the requirements a center must meet for customers rather than speed alone (International Organization for Standardization). That is the right framing. Take the seconds out of the process and the agent keeps the seconds that matter.
Average handle time belongs in your capacity model, not on your agent scorecard. Keep measuring it, keep using it to build schedules and size teams, and stop setting it as a target for individuals.
One change is worth making this quarter: pull handle time alongside first call resolution for each agent, and look at the two together. The agents with low handle time and low resolution are the ones the metric has been rewarding, and that pattern is costing you more contacts than it saves.
Frequently asked questions
What is average handle time?+
Average handle time is the average total time an agent spends on a contact, calculated as talk time plus hold time plus after-contact work, divided by the number of contacts. It is primarily a capacity planning input: multiplied by forecast volume, it produces the workload figure used to determine required staffing by interval.
How do you calculate average handle time?+
Add total talk time, total hold time, and total after-contact work across a set of contacts, then divide by the number of contacts in that set. Calculate it by contact type rather than only in aggregate, since handle time varies substantially by contact driver and a blended figure can hide that variation.
What is a good average handle time?+
There is no universal benchmark worth managing to, because handle time depends almost entirely on contact complexity and mix. A program handling order status and one handling benefits eligibility will differ by minutes for legitimate reasons. Compare an agent only against peers handling a similar contact mix, and read the variation rather than the average.
How do you reduce average handle time without hurting quality?+
Remove work rather than adding speed pressure. The reducible components are usually structural: after-contact work that can be automated, hold time spent on lookups that should be on screen, navigation between disconnected systems, and context lost in transfers. Each of these lowers handle time without asking an agent to rush a customer.
Should average handle time be an agent KPI?+
It works poorly as an individual target, because the easiest ways to lower it are ending contacts early, skipping confirmation steps, and transferring resolvable work, all of which increase repeat contacts. Measure agents on first call resolution, quality score, customer satisfaction, and schedule adherence, and use handle time as diagnostic context.
