Average Resolution Time
Definition
Average Resolution Time
Average resolution time is the mean elapsed time from the moment a customer raises an issue to the moment that issue is fully closed. It counts the clock on the wall, not the minutes an agent spends typing. Those two numbers are not the same.
That gap is where most support reporting goes wrong. A ticket left open for three days may have consumed twenty minutes of real agent work — the elapsed clock and the effort clock measure different things.
Resolution time answers a customer question: how long did I wait? Handle time answers an operations question: how much labour did this cost? You need both, and you need to say which one you are quoting.
Key takeaways
- Average resolution time measures elapsed time to closure, not agent effort.
- Always name the clock: total elapsed, business hours only, or agent-touch minutes.
- Pause states may be excluded, but the exclusion rule has to be disclosed.
- The mean hides long-running outliers, so publish the median and 90th percentile too.
How it works
Start the clock when the customer raises the issue, stop it when the issue is closed for good, then average across every ticket in the period. Everything difficult about this metric lives between those two timestamps.
Three different clocks compete for the same label, and they return wildly different numbers on identical tickets.
| Clock | What it counts | Best used for |
|---|---|---|
| Total elapsed | Every minute from open to close, nights and weekends included | Customer-facing promises |
| Business hours only | Time inside your published support hours | Comparing teams on different shifts |
| Agent-touch time | Minutes an agent actively worked the case | Costing, staffing and capacity plans |
A support contract that promises resolution “within 24 hours” stays unenforceable until it names one of those rows. Two vendors can both hit that target and still deliver completely different experiences.
Pause states are the next argument. Most ticketing tools can stop the clock while you wait on the customer or on a third party — a courier, say, or an upstream software vendor.
Excluding that wait is defensible. Your team genuinely cannot act on a ticket nobody has replied to, and punishing agents for a customer’s silence teaches them to close cases early.
What is not defensible is hiding the rule. Publish the pause definition beside the number, and report paused hours as their own line so a stalled queue never reads as a fast one.
First contact performance drives the whole distribution. When a call center closes an issue on the first interaction, resolution time collapses toward handle time because no queue sits in between.
Every call transfer and every reopen stretches the tail instead. A case that escalates to tier two, waits on a specialist, then reopens a week later can log a hundred elapsed hours on ten minutes of work.
That makes first call resolution (FCR) the strongest influence you have on this number. Fix the transfer rate and the average usually falls without anyone working faster.
Then question the average itself — a small number of very long cases drags the mean upward, so a team with a healthy middle can still post an ugly headline figure.
Report the median alongside it, and add the 90th percentile to expose the tail. If your mean sits well above your median, you have an outlier problem — not a speed problem.
Cost the metric on touch minutes only. The US Bureau of Labor Statistics priced private industry compensation at $46.60 per hour worked in March 2026.
Wages made up $32.60 of that hour, or 69.9 percent, with benefits at $14.01, or 30.1 percent. Multiply that rate by agent-touch time, never by elapsed days.
A ticket resting over a weekend burns no payroll at all. The customer clock keeps running anyway, which is exactly why the two measures need separate reporting lines.
Examples
Resolution time behaves differently in every support setting, so the useful examples are the ones where the clock choice changes the answer. Here are three familiar patterns, each producing more than one defensible number.
Retail banking, disputed transaction. The customer files the claim at 4pm on Friday. An agent spends eleven minutes gathering evidence, then the case waits on the card network until Tuesday morning.
Total elapsed time reads about 90 hours. Business-hours-only reads roughly 16. Agent-touch time reads eleven minutes. All three are accurate — only one belongs in a customer promise.
Software support, reproducible bug. Tier one triages in six minutes and escalates. Engineering ships the fix in the next release, eight days later, and only then does the ticket close.
Here the elapsed average is really measuring release cadence. Reporting it as support performance blames the wrong team, which is why mature groups split product-blocked tickets into their own bucket.
US federal service delivery. The US federal customer experience programme treats waiting time as a matter of public accountability rather than an internal scorecard line.
Executive Order 14058, issued in 2021, established ongoing accountability through High Impact Service Providers and directed 17 agencies to take 36 specific actions to improve customer experience.
Office of Management and Budget (OMB) Circular A-11, Part 6, Section 280 sets the annual guidance, while the 2018 21st Century Integrated Digital Experience Act (IDEA) pushes agencies to modernise digital services.
Staffing shapes all three cases. The US Bureau of Labor Statistics reports a median hourly wage of $20.59 for customer service representatives in May 2024.
It also projects employment declining 5 percent from 2024 to 2034, with about 341,700 openings each year on average as workers leave the occupation. Thinner benches mean longer queues.
Related terms
Average resolution time only makes sense next to the metrics it trades against. These five terms show up in almost every support scorecard, and each one explains a different part of why your resolution clock reads the way it does.
- First Contact Resolution: share of issues closed in a single interaction, and the biggest driver of a short resolution clock.
- Average Handle Time (AHT): agent-touch minutes per contact, routinely confused with elapsed resolution time.
- Service Level Agreement (SLA): contract that has to name which clock the resolution target is measured on.
- Customer Effort Score: survey measure of how hard the fix felt, which long waits reliably push upward.
- Customer Satisfaction (CSAT): post-contact rating that tends to track resolution speed more closely than agent speed.
FAQ
Is average resolution time the same as average handle time?
No. Resolution time measures elapsed time from raise to close, while handle time measures the minutes an agent actively spends on the contact. A three-day ticket can hold twenty minutes of handle time.
Which clock should a service agreement use?
Pick the clock your customers actually experience, then write it into the contract in plain words. Business-hours-only is fair to teams without round-the-clock cover, provided the published hours are easy to find.
Should paused time be excluded from the average?
Excluding waits on the customer or a third party is reasonable, because your team cannot progress the case. Just disclose the rule and report paused hours as a separate line.
Why report the median and 90th percentile as well?
Because the mean gets skewed by a handful of very long cases. The median shows the typical experience, and the 90th percentile shows how bad the tail gets before you intervene.
What counts as a good average resolution time?
There is no universal benchmark, so compare against your own trend and your contracted target rather than an industry figure.
Talk to Outsource Accelerator if you want a support partner who reports the honest clock rather than the flattering one.
Independent
