Average Time on Task
Definition
Average Time on Task
Average time on task is the mean time a worker spends actively doing one defined unit of work: a call, an invoice, a claim, a record. The maths is easy. Deciding what the clock covers, and what it leaves out, is the hard part.
The unit is whatever you define it to be. A contact centre counts calls. A finance team counts invoices.
A claims team counts files. Same formula, different denominator, and a very different conversation with the client about price.
Most teams pull the raw timings from productivity software or a queue log, then pass them to workforce management for the staffing plan.
Task time is the input everything downstream inherits. Get the scope rule wrong and the error travels into your rosters, your quotes and your margin.
Key takeaways
- Average time on task is total active task time divided by the units completed in the same period.
- The scope rule decides the number: hands-on-keys only, or the reading, waiting and switching either side.
- It is a duration per unit, not a share of a shift, which is what separates it from occupancy.
- Per-item pricing in transactional outsourcing runs on this measure, so define it before anyone signs.
How it works
Average time on task divides total active task time by the number of completed units over the same period. The answer is only as honest as the scope rule behind it, so write that rule down before you publish a single figure.
Start with scope. A narrow rule counts only hands-on-keys execution — the seconds between opening the item and closing it. A broad rule adds the reading, waiting and switching either side.
Neither is wrong. A narrow clock flatters the floor. A broad clock tells you what a unit really costs — gaps between items are still time somebody is paid for.
Write the rule as four decisions:
- Name the unit. One call, one invoice, one claim, one record.
- Set the start and stop events. Opening the item and posting it, not logging in and logging out.
- Decide the exclusions. Breaks, training, system outages, coaching.
- Record the complexity mix behind the average, not just the average itself.
Then keep the measure in its own lane. Time per unit answers a different question from the ratio measures around it, and the table below is worth pinning above a reporting pack.
| Measure | What the clock counts | Unit of the answer |
|---|---|---|
| Average time on task | active work on one defined item | minutes per item |
| Average handle time | a whole interaction: talk, hold and wrap | minutes per contact |
| Wrap-up time | after-contact admin only | minutes per contact |
| Occupancy | task time as a share of logged-in time | percentage of a shift |
Handle time is the contact-centre cousin. It measures one whole interaction end to end, so it only works where the unit of work is a conversation.
Task time is broader. It survives in a back office where nobody speaks to a customer at all, which is why finance, claims and records teams reach for it first.
Occupancy is a ratio, not a duration. It compares task time against logged-in available time, so it can climb while time per item stays completely flat.
That difference matters when someone reports an efficiency gain. A rising ratio can mean shorter tasks, or it can simply mean fewer people rostered against the same volume.
Paid breaks sit inside paid hours but outside task time.
Under 29 CFR 785.18, rest periods running from 5 minutes to about 20 minutes are customarily paid and must be counted as hours worked.
So one shift produces two legitimate totals: hours you pay for, and minutes you can attribute to units. Keep both, and never quietly swap one for the other.
Two forces pull the average in opposite directions. Practice pulls it down as people learn the work. Complexity pushes it up as harder items arrive in the queue.
Publish the mix alongside the number — simple items, standard items, exceptions, with a count for each. A moving average with no mix behind it will mislead every reader it reaches.
Examples
Time per unit shows up wherever work is countable. The examples below cover a contact centre, a back-office finance queue and a public-sector service programme, because the same measure carries very different weight in each setting.
Contact centres time each contact type separately, because a password reset and a billing dispute are not the same unit of work.
The US Bureau of Labor Statistics puts the median hourly wage for customer service representatives at $20.59 in May 2024.
That wage is what a minute of task time costs before overheads. The same source projects employment to decline 5 percent from 2024 to 2034, with about 341,700 openings each year on average over the decade.
Fewer seats means the minutes you do buy have to go further — which is the whole argument for measuring time per unit instead of counting heads on a floor.
A finance team processing supplier invoices treats one invoice as one unit. Time it from open to post. Multiply by volume and you have a capacity plan that survives contact with a Monday backlog.
That number is also the price. Transactional outsourcing quotes per item because the buyer can check the arithmetic: task time, times loaded hourly cost, plus margin and an allowance for the gaps.
Per-seat pricing hides all of that. Per-item pricing puts the provider’s task time on the invoice, which is why good providers guard their scope rule so carefully.
Claims teams run the same measure per file, then split it by claim type. A simple motor claim and a disputed liability claim share a queue, not a clock.
Public services measure this way too. The US federal customer experience programme tracks service delivery through designated High Impact Service Providers.
Executive Order 14058, signed in 2021, established ongoing accountability for federal service delivery and directed 17 agencies to take 36 specific actions to improve customer experience.
OMB Circular A-11, Part 6, Section 280 sets the annual federal guidance behind that work. Behind every listed action sits a queue with a clock running on it.
Related terms
Average time on task sits in a family of productivity measures that get mixed up constantly. Each one below answers a different question — so keep them separate in reporting rather than folding them into a single headline number.
- Average Handle Time (AHT): mean duration of a whole customer interaction, talk and hold and wrap combined.
- Wrap-Up Time: after-contact admin minutes that sit inside handle time but outside talk time.
- Idle Time: paid time sitting on the clock with no task attached to it.
- Percent Agent Utilization: share of paid hours a worker spends on productive work rather than waiting.
- Occupancy Rate: share of logged-in time spent handling work instead of sitting available.
FAQ
What is average time on task?
It is the mean time a worker spends actively completing one defined unit of work, such as a call, an invoice or a claim. You get it by dividing total task time by the units completed in the same period.
How is average time on task different from handle time?
Handle time measures one whole customer interaction, so it only applies where the unit of work is a conversation. Task time applies to any countable unit, including back-office items nobody ever speaks about.
Do paid breaks count as task time?
No. Paid rest periods sit inside paid hours but outside task time, so a shift produces one total for payroll and a smaller total for units of work.
Why does average time on task keep changing?
Practice pulls it down as people learn, and complexity pushes it up as harder items reach the queue. Without a published complexity mix, you cannot tell which force moved the number.
Can average time on task be used to price outsourced work?
Yes, it is the unit that lets transactional work be priced per item rather than per seat.
Compare how delivery partners define, measure and price task time across the Outsource Accelerator hubs.
Independent
