Average Handle Time Rporting/Forecasting | Community
Skip to main content

Average Handle Time Rporting/Forecasting

  • November 7, 2025
  • 1 reply
  • 14 views
B

We’re hoping to gain insight into how other customers are forecasting handle time, particularly when using Intercom data. We've noticed discrepancies in the handle time metrics reported by Intercom that don’t align with the structure of our operations.

Typically, I export the raw data and filter out outliers manually. However, we continue to see a significant number of conversations with unusually long handle times that lack a clear explanation. This inconsistency makes it difficult to trust the accuracy of the reports.

We’re looking for a better approach—either a clearer understanding of how Intercom calculates handle time behind the scenes, or alternative methods for forecasting handle time more reliably using Intercom data.

Thank you in advance for any guidance or shared experiences!

1 reply

Emilygav
Intercom Team
Forum|alt.badge.img+6
  • EmilygavIntercom Team
  • Intercom Team
  • November 28, 2025

Hey ​@Bradley Coyle, Emily here from Support Engineering at Intercom 👋🏼

 

Thanks for sharing this - the long handle times you’re seeing typically come from how Intercom measures “handling time,” especially when conversations sit assigned but idle.

 

How Intercom’s handle time works

Handling time includes all open + assigned time until a conversation is closed. It excludes time unassigned, snoozed, closed, or handled by bots. Because of this, if a conversation is unsnoozed or reopened and sits untouched, that idle period still counts and can inflate the metric. Reassignments and long-lived threads can also make totals look unexpectedly large.

 

A better metric to use

For forecasting or performance planning, use Adjusted conversation handling time (or adjusted teammate handling time). These metrics remove the idle “ready but not worked” gaps, so they’re a much closer reflection of true active work.

 

How to get more reliable AHT

  • Use the adjusted metrics in a custom report.

  • Segment by channel and topic — AHT varies a lot across both.

  • Look at medians or P80/P90 instead of raw averages to reduce the impact of long tails.

  • Make sure your office-hours setting matches how you staff and report.

  • If you use tickets, use “time in state” to understand where the work actually happens.

Hopefully this helps explain things for you Bradley!

Terms & ConditionsAccessibility statement