Average handle time explained: what AHT measures and what it is for
Average handle time (AHT) is the mean duration of a customer contact from the moment an agent takes it to the moment after-call work is finished, covering talk time, hold time and wrap-up. It is the primary input to every staffing model in a contact centre, which is what it is genuinely good for.
Also written average handling time, and shortened to AHT in almost every report you will see. In ticketing tools the same idea appears as time to close or agent touch time, which are not the same measurement and should not be compared to a voice AHT.
How is average handle time calculated?
AHT equals total talk time plus total hold time plus total after-call work, divided by the number of contacts handled in the period.
The three components are worth separating in reporting, because they point at completely different fixes:
- Talk time. The conversation itself. Long talk time is usually a complexity or a knowledge problem, not a discipline problem.
- Hold time. The agent looking something up, waiting for a system, or asking a colleague. This is almost always a tooling or access failure, and it is the component most worth attacking.
- After-call work. Notes, dispositions, follow-up tasks. Squeeze it and you damage the case record, which damages the next agent's handle time.
Queue time and abandoned contacts sit outside AHT. They belong to service level, and mixing them in produces a number nobody can act on.
What is a good average handle time?
Call Centre Helper's industry standards give 6 minutes 3 seconds as the general AHT standard, based on 190,702 entries into its Erlang Calculator. The same page shows how little the blended figure is worth on its own: telecommunications averages around 528 seconds, while retail, business and IT services and financial services sit between roughly 282 and 324 seconds.
Use the benchmark to sanity-check your own trend, never to compare two operations. Your six minutes and a competitor's six minutes are different contact mixes, different channel splits and different compliance scripts.
Average handle time: a worked example
A team of 40 agents handles 9,000 contacts a week at an AHT of 7 minutes 30 seconds. That is 1,125 agent hours. Take 30 seconds off and you free 75 hours a week, roughly two full-time positions. Now assume the rushing pushes 4 percent of those contacts into a repeat: 360 extra contacts at 7 minutes 30 costs 45 hours back. The real saving is 30 hours, and the operation has bought it with 360 customers who had to ask twice.
Why AHT and first contact resolution pull against each other
AHT and first contact resolution are the two numbers support teams are measured on, and they are in direct conflict inside every individual contact. The extra 40 seconds an agent spends confirming the fix actually worked, or dealing with the obvious next question before it becomes its own ticket, lands entirely in AHT and pays off entirely in FCR.
Because AHT is measured today and the repeat contact arrives next week, the incentive runs the wrong way by default. An agent who protects handle time is rewarded this month and creates demand that a different agent absorbs in a different reporting period.
The cost asymmetry decides the argument. A saved 40 seconds against a six-minute repeat contact plus queue time is a 1-to-9 trade against you, and it repeats at volume. This is why mature operations report the pair together, target FCR, and treat AHT as a forecasting input and a diagnostic rather than a goal.
How average handle time gets gamed
- Parking the wrap-up. After-call work gets done in an available or unavailable state instead of the wrap state. The work still happens. It just stops being counted.
- Transferring to stop the clock. A transfer ends one contact and starts another, so one long interaction becomes two comfortable ones and the customer explains the problem twice.
- Mute instead of hold. Where the platform excludes hold from the calculation or reports it separately, mute quietly moves time into a friendlier bucket.
- Cherry-picking. Where agents can see or select contact types, the short ones get taken and the hard ones age in queue.
- Blending channels. A chat agent running three concurrent conversations has a handle time that is not comparable to a voice contact. Averaging them produces a number that describes nothing.
- Reporting the mean on a skewed distribution. Contact durations have a long tail. The median and the 90th percentile tell you where the pain is; the mean mostly tells you the tail exists.
What actually reduces handle time without breaking anything
The reductions that hold are the ones that remove work rather than compress it. Hold time is the honest place to start, since most of it is the agent hunting for information the business already has.
- A knowledge base agents actually use. The Consortium for Service Innovation reports that organisations adopting Knowledge-Centered Service see resolution times improve by 25 to 50 percent in the first three to nine months. Practical detail in knowledge base.
- Fewer systems per contact. Every additional login is hold time with a different name.
- Better information arriving with the contact. An agent who can see the problem spends less time constructing it from a description.
- Routing on issue type, not availability. The wrong skill on a complex contact inflates talk time and usually fails resolution too.
One caution on the self-service route. Pushing simple contacts out of the queue raises AHT, because the easy volume that was holding the average down has gone. That is a healthy movement wrongly read as a decline every quarter, and it is a reason to be careful about what ticket deflection is doing to the rest of your reporting. The field-side equivalent of the same trade-off shows up in first-time fix rate, where a rushed visit costs a second one.