Average Handle Time: How to Measure and Reduce It

Average handle time (AHT) measures how much agent effort a single interaction consumes. It’s one of the oldest metrics in support, one of the most useful for planning — and one of the most damaging when used as a target.

How to calculate it

AHT = (total talk/chat time + total hold/research time + total after-contact work)
      ÷ number of contacts handled

The third component is the one teams forget. After-contact work — logging the ticket, updating the record, sending a follow-up, notifying another team — can be a third of the total. AHT calculated without it consistently understates real effort and produces staffing plans that don’t work.

What AHT is genuinely for

Capacity planning. This is the legitimate use. If you handle 2,000 contacts a month at 10 minutes each, that’s roughly 333 agent-hours — about two full-time people before accounting for breaks, meetings and admin. You cannot staff a support team sensibly without this number.

Spotting process problems. A category with an unusually high AHT is telling you something: information is missing at intake, the answer requires checking three systems, or agents are guessing at a policy nobody documented.

Comparing over time within your own team. Industry benchmarks are close to useless because they average across wildly different products. Your own trend is the meaningful comparison.

Rough benchmarks by channel

Channel	Typical AHT	Note
Phone	6–8 min	Concurrency of 1
Live chat	8–12 min agent time	Agents handle 2–3 at once
Email	10–15 min	Includes research and writing
Social	5–10 min	Usually shorter, more public

Chat is the interesting case: AHT per conversation is high, but concurrency means cost per contact is low. An agent handling three chats over 30 minutes has an AHT of 30 minutes each and has resolved three contacts. Comparing chat AHT against phone AHT without adjusting for concurrency will lead you to the wrong conclusion.

Why targeting AHT backfires

This is the central problem with the metric, and it’s worth being blunt about.

AHT is trivially easy to improve by giving worse answers. Reply faster, investigate less, close sooner, skip the verification step. Every one of those reduces AHT immediately.

The costs show up elsewhere and later:

• Reopen rate rises — tickets closed before they were resolved come back.
• Repeat contact rate rises — the customer asks again, so you’ve handled two contacts instead of one. Total effort went up while AHT went down.
• CSAT falls — rushed replies read as rushed.

Teams that put AHT on an individual scorecard reliably get this outcome. Not because agents are cynical, but because that’s what happens when you measure one half of a trade-off.

The rule: never report AHT without reopen rate and CSAT beside it.

When rising AHT is good news

Counter-intuitively, successful deflection makes AHT worse.

If an AI chatbot resolves your simple, fast questions — opening hours, order status, password resets — the contacts remaining for humans are the complex ones. AHT rises, sometimes sharply, while total cost falls and customers are better served.

A team seeing AHT jump after deploying a bot hasn’t got slower. It has stopped counting the two-minute contacts. Judge the change alongside tickets per customer and total agent hours, not in isolation.

What actually reduces AHT safely

These reduce effort without reducing quality:

Better intake. Most handle time is spent establishing what the problem is. An order number, browser and error message collected at intake removes an entire round-trip. See custom forms and support ticket templates.

Customer context in the ticket. If the agent can see plan, order history and past tickets without switching tools, they start informed. This is the main practical argument for CRM ticketing.

Macros for common replies. One click to apply a standard answer, set status and assign. Our canned responses guide covers writing ones that still feel human.

Documentation agents can link to. Retyping an explanation takes four minutes; linking to it takes ten seconds. See how to build a knowledge base.

Fixing the source. If a category has high AHT because the product is confusing, the durable fix is upstream. Support can’t optimise its way out of a product problem.

Better routing. A ticket that reaches the right person first time avoids the reassignment and re-reading that inflate handle time. See ticket routing.

Using it well

1. Report it monthly, not weekly — it’s noisy at small volumes.
2. Segment by category and channel. A blended figure across chat and email tells you nothing actionable.
3. Pair it with reopen rate and CSAT, always.
4. Use it for planning, not performance. Capacity modelling, not scorecards.
5. Investigate outliers rather than the average. The categories at 40 minutes are where the process problems are.

Where EasyChatDesk fits

EasyChatDesk reduces handle time structurally rather than by pressure: custom forms collect the right details up front, CRM ticketing shows customer and order context in the ticket, macros apply canned replies in one click, and the AI chatbot removes the short repetitive contacts entirely.

Pricing is $17/agent/month with reporting on response, resolution and volume included. There’s a 15-day free trial.

The takeaway

AHT is a planning metric that gets misused as a performance metric. Measure it, segment it, use it to size your team and find broken processes — and never put it on a scorecard without reopen rate sitting next to it.

Related: help desk metrics, customer support metrics, and how to improve customer response time.