Open to resolved, with all the waiting counted — the metric that measures the customer’s whole experience of an issue.
Resolution time. Resolution time is the total elapsed time between a support conversation being opened and being fully resolved, including every wait in between. Unlike handle time, which counts only active work, resolution time counts the customer’s whole experience of the issue — queues, hand-offs, nights and waiting for replies included.
The clock starts when the conversation is created and stops when it is marked resolved. Everything between the two decides what the number means, and tools differ on all of it: whether the clock pauses while the conversation waits on the customer, whether it runs on calendar hours or business hours, and what a reopen does — restart the clock, extend it, or spawn a new conversation with a fresh one.
That is the non-obvious point about resolution time: it is not one metric but a family of metrics sharing a name. Two teams reporting “12-hour average resolution” under different pause and reopen policies are not comparable, and neither are two periods of your own history if the policy changed in between. Write the policy down — what pauses the clock, what a reopen does, which hours count — before you trend or compare anything.
Resolution time is the metric closest to how customers actually experience support. A customer never feels your handle time or your first response time in isolation; they feel the total distance between “I have a problem” and “it’s fixed.” It is also the metric where process problems surface: a long tail of slow resolutions almost always traces to hand-offs, stalled escalations, or conversations waiting on a third party — things no amount of faster typing fixes.
Resolution time calculation. Resolution time = resolved-at timestamp − created-at timestamp, minus any periods the conversation waited on the customer if your policy pauses the clock, reported as a median across conversations. Track the 90th percentile alongside the median to keep the slowest cases visible.
Use the median, not the mean: resolution times are heavily skewed, and one conversation stuck for three weeks can move an average more than a hundred routine ones. Then read the tail deliberately — sort the slowest ten percent by cause, not by agent. If you pause the clock for waiting-on-customer, audit that state occasionally; it is the easiest place for a team to quietly park hard conversations, and a paused clock that never resumes is a customer who gave up.
Give your customers faster answers and your team their evenings back.
Desktop and mobile. Free for solo.