Average handle time
In one sentence
Average handle time is the mean duration of a customer interaction, including talk time, hold time and the after-call work done once the customer has gone; it is a cost driver, not a measure of quality.
Definition
Average handle time is the mean length of a customer interaction, counting talk time, hold time and the wrap-up work done immediately after the customer has gone.
It is the classic contact center efficiency metric and the most frequently misused one, because a lower number can mean either a tighter process or a customer who was cut off before the problem was solved.
What it comprises
- Talk time.
- Hold time.
- After-call work, meaning the wrap-up performed once the customer has left the interaction.
- Excluding after-call work understates the true cost of an interaction, which is the most common way the number is quietly flattered.
Why optimizing it directly is dangerous
- The fastest way to reduce handle time is to end interactions before problems are solved.
- That shifts cost into repeat contacts, which is invisible in the handle time report and expensive everywhere else.
- Handle time and first contact resolution pull against each other, and the pairing is the only honest way to read either one.
The legitimate uses
- Capacity planning and staffing, where it is genuinely necessary.
- Detecting process problems, where a rising trend on one issue type points to friction.
- Comparing interaction types to identify automation candidates.
Where AI changes the picture
- Automating simple, short interactions raises human handle time, because the remaining human workload is the complex residue.
- A rising AHT after deploying automation is usually a success signal misread as a failure. This misreading is common enough to be worth stating plainly.
The on-site analogue
- Conversation duration on a website voice agent behaves differently. On a commerce site, a longer conversation frequently signals deeper engagement rather than difficulty.
- Duration should be read alongside outcome rather than minimized on its own.
Common misconception
That a lower handle time indicates better service. It indicates shorter interactions, which may be efficiency or may be abandonment. The number alone cannot tell you which, and read on its own it rewards the wrong one.
Why it matters commercially
Handle time drives staffing cost and appears in every operational report, so it is the number finance and workforce management reach for first. Understanding its interaction with resolution is what separates useful reporting from harmful targets: set it without a resolution guardrail and you have paid people to end contacts early.
In voice specifically
On a website, conversation duration is not a cost to be minimized in the same way. A longer voice conversation on a commerce page often means a more engaged visitor, so duration should be read next to the outcome, whether the answer landed and whether the visitor converted, rather than pushed down for its own sake.
Where AsqVox fits
Conversation duration contributes to per-minute cost, so shorter answers are better both economically and experientially. That alignment is deliberate: on AsqVox the incentive to answer efficiently and the incentive to answer well point the same way, which is exactly what handle time targets break in a contact center.
Visual
The metric that gets worse when things get better
Human handle time rising after automation is usually a success signal. The easy calls left; the hard ones remain.
Statistics
Every figure carries its source and year. Vendor numbers are labelled as vendor numbers, and where no reliable figure exists this page says so rather than borrowing one.
A call handled by a person costs roughly USD 7 to USD 12, and handle time is the primary driver of that figure.
USD 7 to 12industry rangeWidely cited industry range, 2025 - Not an audited figure. Because handle time is what drives the human cost, it is the lever finance watches, and the reason a resolution guardrail matters when it becomes a target.
A call handled by a voice agent costs roughly USD 0.40.
USD 0.40industry rangeWidely cited industry range, 2025 - Same caveat. Treat the ratio between the human and agent figures as directional rather than treating either absolute number as precise.
All-in production voice agent cost lands around USD 0.11 to USD 0.33 per minute, which makes duration a direct cost input for automated interactions too.
USD 0.11 to 0.33 per minuteindustry rangeIndustry-reported range, 2025 - For an automated channel, duration is not a proxy for cost, it is the cost. That is why a shorter correct answer is straightforwardly better on a per-minute model, with none of the resolution trade-off a human handle time target carries.
Conversational AI in contact centers was forecast to cut agent labor costs by USD 80 billion by 2026, with one in ten agent interactions automated by 2026.
USD 80bnanalyst forecastGartner press release, attributed to VP analyst Daniel O'Connell, 2022 - Dated 31 August 2022, and the date belongs in any citation. The one-in-ten figure is the informative half: even a modest automation share moves the human handle time mix, because it is the simple calls that leave first.
There is no reliable current cross-industry AHT benchmark, and comparison across organizations is meaningless given differing interaction mixes.
-no reliable figureTwo centers with different products, customers and channels will have different handle times for reasons that have nothing to do with quality. Internal trend on a stable interaction mix is the only defensible use.
There is no published data quantifying the AHT increase attributable to automating simple interactions, despite the effect being well recognized in practice.
-no reliable figureThe direction is not in dispute: automating the short easy calls raises the average length of what remains. The magnitude is not published, so the worked example on this page uses an illustrative rise of about a third rather than a sourced figure.
Examples
In practice
An organization deploys voice automation and human AHT rises by roughly a third. The operations report flags it as a degradation. Analysis of the interaction mix shows short informational calls have moved to the agent, leaving complex cases with humans. Total cost per resolved issue fell substantially. The metric moved in the wrong direction for the right reason, and reading it beside resolution is what caught that.
The everyday version
Average handle time is how long a typical customer interaction takes. It is useful for working out how many people you need. It is a bad target, because the quickest way to shorten calls is to hang up before the problem is solved, which just means the customer rings back tomorrow and you pay for the call twice.
Usage
Who says it
- Contact center operations and workforce management, constantly, as the core efficiency number.
- Finance, inside staffing cost models, where handle time sets headcount.
Where it turns up
- Next to staffing models, containment, first contact resolution and cost per contact.
- In operational dashboards and RFPs, where the useful question is not the target but whether a resolution guardrail sits beside it.
Common misuse
- Setting it as a target with no resolution guardrail, which pays agents to end contacts early.
- Excluding after-call work, which understates the true cost of an interaction.
- Reading a post-automation increase as a failure, when it is usually the expected result of moving simple calls off the humans.
Questions people ask
What does average handle time include?
Talk time, hold time and after-call work, the wrap-up done once the customer has gone. Excluding after-call work is the most common way the number gets flattered, because it hides real cost. AHT is the mean of that total across interactions.
Why is average handle time a bad target?
Because the fastest way to reduce it is to end interactions before the problem is solved. That does not remove the cost, it moves it into repeat contacts, which are invisible in the handle time report and expensive everywhere else. AHT and first contact resolution pull against each other, so a handle time target without a resolution guardrail rewards the wrong behavior.
Why did our AHT rise after we deployed automation?
Almost certainly because it worked. Automating the short, simple interactions leaves the complex residue with your people, so the average length of a human-handled call goes up. Total cost per resolved issue usually falls at the same time. A rising post-automation AHT is a success signal that is routinely misread as a failure.
What is the difference between average handle time and after-call work?
After-call work is one of the three components of average handle time, alongside talk time and hold time. It is the wrap-up performed after the customer has left. AHT is the whole interaction; after-call work is the part that is easiest to leave out, and leaving it out understates the true cost.
Last reviewed 4 August 2026. Written and reviewed by Dhruv Dholakia, founder of AsqVox.