Low AHT Shouldn’t Be the Only Goal

Some metrics gradually stop being measurement tools and become the goal itself. Average Handle Time (AHT) — also called Average Call Handling Time (ACHT) — is the best-known example. It measures the average time spent on a call.

Once an operation sets an AHT target of 360 seconds, a simple rule of thumb takes hold very quickly: under 360 is good, over 360 is bad.

For management this is very convenient. The number is clear, easy to measure and comparable across teams. But there are things those 360 seconds don’t show: what was the customer’s problem, how complex was the conversation, was the issue actually resolved, did the customer call back two days later about the same thing?

So perhaps the question we should be asking is this:

Not “How do we lower AHT?” but “What is the right AHT for our operation?”

Why AHT is indispensable

Let’s clear up one misunderstanding first: AHT is not a bad KPI. It plays a direct role in forecasting, capacity planning, service level, shift requirements and FTE calculations.

Consider the impact of a change that looks small. In an operation handling 10,000 calls a day, if AHT rises from 400 to 450 seconds:

10,000 calls × 50 seconds = 500,000 seconds ≈ 139 hours of additional workload.

Even before shrinkage and occupancy are taken into account, that means roughly 17 extra eight-hour shifts a day. From a planning perspective, 50 seconds is not a small number at all.

So the problem isn’t that AHT is tracked. The problem is when a metric designed for capacity planning is turned directly into an individual performance metric. AHT is a powerful tool for planning an operation; on its own, it is quite limited for understanding how well an agent is doing.

Is lower AHT really better performance?

Let’s compare two agents:

Looking at AHT alone, A is clearly more efficient: every call is closed a minute faster.

Now let’s ask a simple question: how much time does each agent spend to actually resolve a customer’s problem? Since every unresolved issue means another call, time per resolution can be roughly calculated as AHT / FCR:

• Agent A: 240 s / 0.60 = 400 seconds
• Agent B: 300 s / 0.90 = 333 seconds

The picture flips. A, who looks faster per call, consumes about 20% more resources per resolution than B. And this calculation is optimistic: in practice, a customer who calls back is usually more frustrated, and the second conversation takes longer than the first.

This relationship is observed in the field as well. SQM Group’s long-running research shows that every 1-point improvement in FCR comes with roughly a 1-point improvement in CSAT.

In short, low AHT is sometimes not real efficiency but cost deferred to the next call.

What we measure changes behavior

Turning AHT into an individual target has a deeper consequence: human behavior.

The principle named after economist Charles Goodhart sums it up well:

When a measure becomes a target, it ceases to be a good measure.

When we tie a KPI to rankings, bonuses or performance reviews, we are effectively telling employees which behavior is valued. If the main success criterion is low AHT, the agent adapts over time: shortens explanations, doesn’t ask whether the customer needs anything else, doesn’t spend time getting information from another department, and tries to end the call at the earliest point instead of explaining the alternatives.

None of this is bad intent. The agent is adapting to the target they were given.

What we measure changes behavior. What we reward makes that behavior permanent.

That is why designing an AHT target is not a reporting matter; it is a decision about what kind of relationship we want to build with the customer.

So is high AHT good?

Of course not. This is the second trap the AHT critique easily falls into.

Saying “low AHT isn’t always good” is not the same as saying “the higher the AHT, the better”. If a call that could be resolved well in four minutes takes six, there are two minutes of real inefficiency. The source of that gap may be slow systems, knowledge gaps, unnecessary hold time, complicated procedures, information scattered across screens or badly designed processes. All of them are operational problems that need fixing.

The aim is not to defend high AHT but to separate time that adds value for the customer from time that creates none.

Thirty seconds spent understanding the customer’s problem and thirty seconds the customer spends waiting for a slow screen look identical in the AHT report. For the operation they are completely different things.

A healthy range instead of a single target

Instead of a single target in seconds, a healthy AHT range approach gives far more meaningful results. The method is simple: split existing calls into AHT bands and compare each band’s CSAT, FCR, repeat-call and quality results.

An example distribution might look like this (figures are illustrative):

The signal is clear: up to 420 seconds, every extra minute given to the customer noticeably increases satisfaction. From 300 to 420, CSAT rises by 12 points. After 420, the return on extra time drops to almost zero.

In this case, a rigid 360-second target pushes agents to speed up precisely in the zone where they are still creating value for the customer. Calls above 500 seconds, meanwhile, deserve a separate look.

What we are looking for is not the minimum AHT but the optimum AHT zone.

Not every call should have the same AHT

Targets based on averages have another problem: they assume every call is equally difficult. Yet customer requests differ considerably in how long they take:

Now imagine two agents on the same team who work at exactly the same speed on every call type. The only difference is the mix of calls they receive:

• Agent X: 70% password resets, 30% complaints → average 319 seconds
• Agent Y: 30% password resets, 70% complaints → average 491 seconds

There is a 172-second gap between them, and all of it comes from the call mix, not from performance. Against a blanket 400-second target, X looks like a star and Y like a problem agent.

That is why, especially in skill-based operations or those with very different call types, the relationship that really matters is:

Call reason × AHT × CSAT × FCR

The question should not be “what is this agent’s AHT?” but “how does their AHT compare with similar calls of this type, and what is the outcome for the customer?”

The average doesn’t tell the whole story

An operation’s average AHT might be 430 seconds. But when you look at the distribution, you might see this:

• Median370 s
• P75450 s
• P90720 s

The 60-second gap between the median and the average shows that the distribution has a long right tail. In other words, there is no AHT problem spread across the whole operation; the problem is that a small share of calls run far too long.

A manager who looks only at the average tells the whole team “your AHT is high”. The right move, though, is not to try to speed up every agent but to look at the calls above P90. Maybe a particular transaction type takes long, maybe a particular screen is problematic, maybe a few agents need coaching — or maybe customers genuinely need more time in that process.

AHT analysis must always show, alongside the average, the median, percentile distributions and call-type breakdowns.

How can AHT be used in a performance system?

Old model: Target 400 seconds. 390 is good, 420 is bad.

Alternative model: Healthy range 380–440 seconds.

An agent at 395 is fine, and so is one at 430. When an agent drops to 290 seconds, the system doesn’t automatically count it as outstanding performance; it looks at CSAT, FCR and quality results. When an agent rises to 470 seconds, it doesn’t automatically count it as failure either; it examines the call types and outcomes.

In this approach AHT stops being a race metric and becomes a control metric. Real performance is judged by how accurately and durably we resolve the customer’s problem.

What should we see on the dashboard?

A screen that just says AHT: 412 s tells you very little. Seeing this set side by side is far more meaningful:

• AHT412 s
• CSAT84%
• FCR89%
• Repeat calls8%
• Quality94%

Two simple charts added underneath can change the view entirely. An AHT–CSAT scatter plot shows how satisfaction changes at different duration levels. An AHT distribution histogram reveals whether the problem lies in the overall average, in specific call types or in the outliers.

We no longer say “AHT is high”. We ask: where, on which calls, and with what outcome is it high? This small change moves KPI management from reporting to analysis.

Conclusion

AHT will never disappear from contact centers. Nor should it. For workforce management, capacity planning and cost management it is still one of the industry’s most important indicators.

But when we detach AHT from the customer experience and from the outcome of the conversation, what we measure stops being efficiency. All that’s left is speed. And speed and efficiency are not the same thing.

A good contact center is not the one with the shortest calls, but the one that resolves the customer’s problem without wasting time — but without skimping on the time it needs.

Don’t minimize AHT. Optimize it.