Skip to content

AHT (Average Handle Time)

The average time it takes to fully resolve a customer conversation, from first reply to closure. A workload metric, not a quality metric.

AHT stands for Average Handle Time. It is the average time it takes a team to fully handle a customer conversation, measured from first reply to final closure. Some teams include time the conversation spent waiting on the customer, most do not. The metric should be defined the same way every month or comparison is meaningless.

It is a workload metric. It tells you how much time conversations consume. It does not tell you whether the customer was happy or whether the answer was right.

Why AHT can mislead

A team that rushes its replies has a low AHT and a low CSAT. A team that writes patient, complete answers has a higher AHT and a higher CSAT. Optimizing for AHT alone makes service worse. Optimizing for AHT alongside CSAT and resolution rate makes it better.

AHT is also distorted by the mix of conversations. A week with more refund requests will have a different AHT than a week with more product questions, even if nothing changed about the team. Watch AHT by category, not just overall.

How AI affects AHT

AI changes AHT in two ways. The first is direct: an AI agent resolving a conversation end-to-end skips human time entirely. Those conversations have an AHT of zero from the team's perspective. They lower the overall average mechanically.

The second is indirect: an AI that drafts replies, summarizes context, and surfaces relevant policy cuts the human's work per conversation. The human still handles the case, but the prep time, the search time, the rewrite cycle, all shrink. Teams using AI assistance well often see human AHT drop twenty to forty percent on the conversations the AI did not fully deflect.

What good looks like

For consumer ecommerce, AHT of five to fifteen minutes is normal. For B2B support with complex tickets, thirty minutes to several hours. For technical support involving log lookups and debugging, longer still. There is no universal target. There is only the trend in your data and the trade-off with CSAT.

In Keloa

In Keloa, AHT is reported alongside deflection rate and CSAT, so you can see whether faster handling is also better handling. The AI agent deflects what it can and assists humans on the rest. See the product for the dashboard.

See how this plays out in the product.

Free Starter plan, 50 AI replies, no credit card. Set up in ten minutes.