The 30-second answer

AHT measures three things added together: talk time (the live conversation), hold time (the customer waiting on hold), and after-call work (the agent typing the summary, picking the disposition code, doing the wrap-up). Add those three for every contact, take the average, that's your AHT.

A 6-minute call with 30 seconds of hold and 90 seconds of wrap-up has an AHT of 8 minutes flat.

Why teams obsess over it

AHT maps directly to staffing. If your AHT drops from 8 minutes to 7, you can either handle 14% more contacts with the same headcount or run with 14% fewer agents. That's a real cost line, and it's the number a CFO can act on.

It's also the number a contact-centre platform makes easiest to report. Every CCaaS (NICE CXone, Genesys Cloud, Five9) shows AHT on the front dashboard. It's the number leaders are trained to optimise.

The honest critique

Compressing AHT often pushes the problem somewhere else. The agent rushes the resolution, the customer doesn't fully understand, and they come back tomorrow — which shows up as a higher recontact rate and a lower CSAT. Total work per customer hasn't gone down; it's just spread across more contacts.

A team that drops AHT by 30 seconds and accepts a 5-point CSAT drop probably hasn't saved anything. Recontact tickets, complaint escalations and quality-coaching time eat the gains.

The honest version: AHT is a real efficiency lever when it goes down for the right reasons (better tooling, better knowledge base, fewer dead-end conversations) and a trap when it goes down for the wrong ones (rushing).

How AI changes AHT

Two distinct effects:

  • Agent-assist compresses human AHT. The agent gets suggested replies, auto-drafted summaries, and surfaced knowledge. Talk time falls; after-call work falls more. Good vendors (Cresta, ASAPP, Balto) report 20-40% AHT reduction on the queues where it lands.
  • Autonomous AI agents remove human AHT entirely. The bot handles the contact without an agent ever joining. AHT for that contact drops to zero on the human side. Whether the customer's problem actually got solved is a separate question — see resolution rate.

How to use AHT well

Pair it with three things that catch the failure modes:

  • CSAT on the same contacts. If AHT drops and CSAT drops, the gain isn't real.
  • Recontact rate within 72 hours. If AHT drops and recontact rises, the work just moved to a second contact.
  • First Contact Resolution. The honest counterweight — a high-FCR, lower-AHT outcome is the only one worth celebrating.

Reported on its own, AHT tells you the contact ended quickly. Reported with the trio above, it tells you whether the customer was actually served. The second is what matters.