What is average handle time (AHT)? Formula, examples, and how to improve it
What is average handle time?
Average handle time, or AHT, measures the average time a contact center spends handling an interaction. It usually includes:
- Talk time between the customer and employee
- Customer hold time during the interaction
- After-call work, also called wrap-up time
AHT is useful for staffing and for finding where work gets stuck. It doesn't show whether the customer got good service. A common mistake: AHT falls because conversations end sooner, while repeat contacts, transfers, or unresolved issues go up. The team looks more efficient, but the work has just moved somewhere the metric doesn't count.
To use AHT well, define exactly what it counts, break it into its parts, and read it next to customer outcomes.
Queue wait before an employee answers is usually measured separately, as are abandoned calls and any work that isn't attached to the contact record.
Why average handle time matters
AHT matters because handle time drives most of a contact center's labor cost. It helps answer four questions:
- How many people do we need? Staffing models turn expected contact volume and AHT into the number of people needed for each shift. A small error in AHT adds up across every hour of the schedule.
- What does a contact cost? Handle time multiplied by loaded labor cost gives a starting point for cost per contact, and for comparing journeys and channels.
- Where is work getting stuck? A change in talk, hold, or after-call work often points to a specific constraint: a slow system, a missing permission, a hard-to-explain policy, or a manual documentation step.
- Which journeys are harder than they should be? When one contact reason consistently takes longer than similar ones, the cause is often in the product, the policy, or the knowledge behind it, not the employee handling it.
None of these questions asks whether the customer got what they needed. That's why AHT works best next to outcome measures, not as a target on its own.
How to calculate AHT
For voice contacts, use this common formula:
AHT = (total talk time + total hold time + total after-call work) / handled contacts
Suppose a queue handled 10 calls with these totals:
| AHT component | Total time |
|---|---|
| Talk time | 4,200 seconds |
| Hold time | 900 seconds |
| After-call work | 900 seconds |
| Total handle time | 6,000 seconds |
AHT = 6,000 seconds / 10 handled calls = 600 seconds, or 10 minutes
For a weighted rollup, add component time and handled contacts across the included groups, then divide. Do not average two team-level AHT values unless the teams handled the same number of contacts.
Weighted AHT example
Team A handled 100 contacts at an AHT of 480 seconds. Team B handled 300 contacts at 600 seconds.
Weighted AHT = ((100 x 480) + (300 x 600)) / 400 = 570 seconds
A simple average of 480 and 600 would produce 540 seconds and understate the larger team’s contribution.
What’s a good average handle time?
A good average handle time is one that is stable for comparable work and paired with strong resolution. No single number is right across industries, channels, or journeys. A password reset and a disputed insurance claim can both be one “contact,” but they should not share a target.
A more useful baseline comes from your own data:
- Segment by contact reason. Set a range for each journey rather than one target for the queue.
- Use your recent history. Take the median and spread for each journey over a stable period, after any platform or process change.
- Check outcomes at each level. Look at resolution and repeat-contact rates for contacts across the range. If the fastest contacts also have the highest repeat rate, the efficient handle time is longer than the fastest one.
- Watch the trend, not the point. A good AHT for a journey is one that moves for reasons you can explain.
Document how you define AHT
Before you compare queues, channels, platforms, or time periods, write down exactly what your AHT includes.
| Field | What to decide | Why it matters |
|---|---|---|
| Start event | When handling begins | Assignment, acceptance, and connection are different timestamps |
| Stop event | When handling ends | Disconnect and completion of after-call work are different |
| Components | Talk, hold, after-call work, pause, active work | Platforms include different states |
| Denominator | Which handled contacts count | Transfers, callbacks, and short contacts can change the average |
| Queue time | Included or reported separately | AHT is not normally a wait-time metric |
| Transfers | One contact, each employee's part of the contact, or both | Employee workload and customer journey time answer different questions |
| Concurrency | How simultaneous digital work is allocated | Wall-clock time can overstate employee workload |
| After-call work | Maximums, auto-close, and reopened records | System rules can move the metric without changing the work |
| Channel | Separate voice, chat, email, messaging, and task definitions | The same clock does not fit every channel |
| Handler | Human, AI, or blended | AI agents and employees work differently, so time each one separately |
| Segments | Journey, queue, outcome, transfer, complexity | Mix shifts can move AHT without process change |
| Owner | Team accountable for definition and review | Prevents silent metric drift |
Keep this definition with the dashboard or metric dictionary. If a platform change alters how time is recorded, start a new baseline instead of joining old and new data into one trend.
How AHT differs by channel
Voice
The standard formula fits voice best, because talk, hold, and after-call work are usually separate states. Still report inbound, outbound, callback, and transferred calls separately when they work differently.
Live chat and messaging
A chat may stay open while the customer reads, steps away, or does something else, and one employee may handle several chats at once. Total elapsed time shows how long the customer spent. Active handling time shows how much work the employee did.
Report both when they support decisions:
- Customer elapsed time: from the defined start to completion
- Active handle time: the work attributed to the employee or system
Do not compare a concurrent-chat AHT directly with voice AHT and conclude that one channel is more efficient.
Email and asynchronous cases
Elapsed time from receipt to closure may include hours when no one is working the case. Measure active handling time for workload and time to resolution for the customer experience. Make assignment, reopen, pending-customer, and after-work rules explicit.
Transferred contacts
A transfer can create several valid clocks:
- Handle time for the first employee
- Handle time for the receiving employee
- Total handling across both employee legs
- Customer elapsed time from request to outcome
Staffing plans may need each employee's time. Service leaders also need the full customer view, so that a quick transfer isn't counted as a quick resolution.
Why lower AHT is not always better
AHT measures how much time work takes, not whether the customer's issue was resolved. A drop in AHT can be a bad sign when it comes from:
- Ending a conversation before the need is resolved
- Avoiding a transfer that the customer actually needs
- Skipping verification, documentation, or policy steps
- Rushing a complex or emotional conversation
- Moving work into follow-up that the contact record does not count
- Discouraging employees from investigating the root problem
Read AHT next to at least one measure from each row below:
| Operating question | Paired measures |
|---|---|
| Did the customer get help? | Verified resolution, first-contact resolution, repeat contact |
| How did the experience feel? | CSAT, customer effort score, complaints |
| Was the work controlled? | Quality evaluation, policy exception, required-step completion |
| What did the outcome cost? | Cost per resolution, transfer rate, rework |
| Can the operation sustain it? | Occupancy, absence, attrition, coaching time |
If AHT falls while first-contact resolution falls, the team may have shortened the contact by deferring the work. If AHT rises while repeat contacts fall and resolution improves, the operation may be spending time more effectively.
How to improve AHT without gaming the metric
Fix what takes the time, not the clock.
- Remove repeated work. Carry verified identity, customer details, and journey context across channels and handoffs.
- Fix system access. Reduce application switching and bring necessary actions into the service workflow.
- Improve knowledge where decisions are made. Make current policies and procedures easy to find and apply, ideally inside the conversation rather than in a separate search tool.
- Reduce avoidable holds and transfers. Clarify authority, routing, and specialist access.
- Automate after-call work carefully. Generate summaries or dispositions with review where the consequence requires it.
- Coach to outcome and quality. Use conversation evidence to address a specific behavior rather than imposing a universal time target.
- Simplify the journey. Remove policy, verification, or data steps that add effort without a clear reason.
- Measure matched work. Compare the same journey and complexity before and after a change.
Use distributions as well as averages. A stable AHT can hide a growing group of very long contacts, while a small number of extreme cases can move the mean without indicating a broad process problem.
Measuring AHT across AI and human handoffs
An AI-enabled service journey needs three distinct clocks:
- AI interaction time: Time the customer spends with the AI agent before resolution or handoff.
- Human handle time after handoff: Talk, hold, and after-contact work for the employee who receives the case.
- Total customer time to verified outcome: Time from the initial request until the need is completed, including transfers, waits, and follow-up.
The first two support capacity and system diagnostics. The third protects the customer outcome.
Also preserve handoff quality. Did the employee receive the customer’s goal, authentication state, relevant context, steps already attempted, and reason for escalation? A shorter employee AHT created by a complete handoff can be valuable. A shorter AI interaction that forces the customer to repeat everything is not.
AI agents built on Sierra resolve many customer requests end to end, on voice and digital channels. When a customer asks for a person, or a request needs an employee's judgment, the handoff can pass along the customer's intent, a summary of the conversation, and the steps already taken. The employee can start on the problem instead of asking the same questions again, which can shorten human handle time without adding customer effort.
In Sierra, Ghostwriter helps teams review resolution, categorize conversations, and audit quality at the conversation level. That shows all three clocks for a journey together, so you can check whether faster handling still ends with the customer's need resolved.
Use AHT to plan and diagnose, not to rush customers
AHT should help leaders forecast workload and find friction. It should not tell employees how quickly every customer deserves to be handled.
In your next operating review, break AHT into its parts, segment it by similar journeys and outcomes, and put resolution and customer effort next to it. If the team can't explain which work got faster and whether customers still got what they needed, a lower average isn't proof of improvement.
