Mean Time to Resolve
Definition
Mean Time to Resolve
Mean time to resolve is the average total time from an incident starting to it being fully resolved, including detection, response, repair, and removal of the underlying cause. It is the customer’s view of an incident, not the engineering team’s.
Restoring service is not resolving it. A workaround that keeps the system running leaves the incident open — in every sense that matters to the person who reported it.
That is why the figure runs longer than repair time. It includes the waiting, the investigation, and the permanent fix that follows.
Key takeaways
- Mean time to resolve averages the full incident lifespan from start to permanent fix.
- It includes detection and waiting time, which repair-only measures exclude.
- Workarounds pause the impact but should not close the incident.
- Recurrence within a defined window means the incident was never resolved.
How it works
Mean time to resolve is calculated by summing the elapsed time from incident start to full resolution across every incident in a period, then dividing by the number of incidents resolved during that period.
The formula is: total resolution time ÷ number of incidents resolved.
Because it spans the whole timeline, the number only becomes actionable once broken into its parts.
| Component | Typical share of total | Owned by |
|---|---|---|
| Detection | Often the largest slice | Monitoring, service desk |
| Response | Small, but variable | On-call rota |
| Repair or workaround | Moderate | Engineering |
| Root cause removal | Long tail | Problem management |
Detection routinely dominates and routinely goes unexamined. Teams optimise the repair step because it is visible — while the blind window before anyone noticed stays untouched.
Set a recurrence window and honour it. An incident that returns inside 30 days should reopen the original record rather than start a new one.
Public services measure the citizen-facing side of this. The U.S. federal customer experience program collects and publishes service-provider feedback on how services actually perform.
Assisted service volumes remain substantial. The IRS Data Book records 50.4 million taxpayers assisted by calling or visiting an IRS office in FY 2025.
Report the median as well as the mean. One three-week incident will drag an otherwise healthy monthly average past any target you set.
The measure overlaps with resolution time in service-desk reporting, and the two names are often used for the same field.
Never close an incident to protect the metric. Premature closure moves the problem into the recurrence column, where it costs more and looks worse.
Examples
Resolution timelines vary by how much investigation a fix requires and whether a workaround is acceptable in the meantime. Five cases show the practical range.
Service desks resolve most password and access tickets within minutes. High volume and a known fix keep the average low and the median lower still.
Software vendors resolve defects across release cycles. Service is restored by workaround while the permanent fix waits for the next scheduled deployment.
Banks resolve payment failures in two stages. The customer is made whole immediately, and the underlying cause is worked separately under problem management.
Manufacturers resolve equipment faults only after root cause is confirmed. Restoring a line without knowing why it stopped invites the same outage next week.
Outsourced service desks contract on resolution with a recurrence clause — buyers should insist that a repeat inside the window reopens the original ticket rather than creating a fresh one.
Related terms
Mean time to resolve closes the incident timeline that detection, response, and repair measures open. The terms below cover the neighbouring metrics, the tooling, and the contracts involved.
- Resolution Time: the service-desk field most often holding this same measure.
- First Contact Resolution: the front-line outcome that keeps incidents out of this measure entirely.
- Support Ticket: the record the clock runs against.
- Ticketing System: the tool supplying start, restore, and close timestamps.
- Escalation: the process that shortens resolution when the first team cannot fix it.
- Help Desk Support: the function that owns communication throughout.
- Service Level Agreement (SLA): the contract that sets resolution targets by severity.
FAQ
How is mean time to resolve calculated?
Sum the elapsed time from incident start to full resolution across all incidents, then divide by the number resolved.
How does it differ from mean time to repair?
Repair stops when service returns, while resolve continues until the underlying cause has been removed.
Does a workaround close the incident?
No. It restores service and should pause customer impact, but the record stays open until the permanent fix lands.
Why report the median too?
Because a single very long incident distorts the mean and hides otherwise healthy performance.
What should the recurrence window be?
Thirty days suits most service desks, with the original record reopening if the issue returns inside it.
Which component is usually slowest?
Detection, which is also the component teams examine least.
Buyers comparing outsourced service-desk partners on resolution performance can review vetted providers in the Outsource Accelerator directory.
Independent
