What MTTR is
MTTR is a mean time metric used in incident reporting, and it expands four different ways: mean time to repair, to restore, to respond and to resolve. A number quoted without its expansion is unusable, which is the first thing to fix in any ITSM reporting pack.
Category: ITSM ToolsOpen ITSM Tools
What MTTR means
MTTR is a mean time metric for how long incidents take, calculated as total elapsed time divided by the number of incidents in the period. The acronym covers four distinct measurements: mean time to repair, mean time to restore (or recovery), mean time to respond and mean time to resolve. They measure different intervals and routinely differ by hours on the same incident set.
The four MTTR measurements
Each expansion starts and stops at a different point, which is why two teams can report honest MTTR figures that are nowhere near each other.
What each MTTR expansion measures
| Expansion | Interval measured |
|---|---|
| Mean time to respond | Alert raised to a human acknowledging it |
| Mean time to repair | Work starting to the fix being applied |
| Mean time to restore | Failure starting to service working again |
| Mean time to resolve | Failure starting to the underlying cause being closed out |
Calculating MTTR
The arithmetic is simple: add the intervals for every incident in the period and divide by the incident count. Twelve incidents consuming 18 hours of restore time give an MTTR of 1.5 hours.
MTBF, mean time between failures, is the matching reliability metric for repairable systems. MTTF, mean time to failure, covers parts you replace rather than repair. Availability follows from two of them: MTBF divided by the sum of MTBF and MTTR, which is why halving restore time moves availability as much as doubling the interval between failures.
Why the mean misleads
MTTR is an average, so one eight-hour outage buried in fifty five-minute tickets disappears into the number, and the number improves every time you log more small tickets. The incidents worth managing are the tail.
Report a percentile alongside it, usually p90 or p95 restore time, and keep the incident count visible so nobody reads a falling mean as improving reliability. If a service target is written against MTTR, make the contract state the expansion, the clock start, and whether planned maintenance and waiting-on-customer time are excluded. Those exclusions are where most SLA disputes originate.
MTTR in a tool evaluation
Treat MTTR reporting as a tier question rather than a feature question. In Zendesk, deeper service level management arrives with Suite Growth at $89 per agent per month billed annually, checked 2026-03-16, above Suite Team at $55. In Jira Service Management the full incident module sits in Premium at $44.27 per agent per month billed annually, checked 2026-03-17.
Monitoring spend hits the same metric from the other end, since detection and diagnosis time are usually the largest share of restore time. The downtime cost calculator puts a figure on an hour of outage, which is the only sane way to judge whether a reporting tier or a monitoring upgrade is worth buying.
MTTR questions
How do I calculate MTTR?
Add the elapsed time for every incident in the period and divide by the number of incidents. Twelve incidents and 18 hours of total restore time give an MTTR of 1.5 hours. State which expansion you used, because repair, restore, respond and resolve measure different intervals.
What is the difference between MTTR, MTBF and MTTF?
MTTR measures how long a failure lasts. MTBF, mean time between failures, measures how long a repairable system runs between them. MTTF, mean time to failure, applies to parts that get replaced rather than repaired. Availability is MTBF divided by MTBF plus MTTR.
Is MTTR a good KPI?
It is useful with guardrails and misleading without them. As a mean it hides the long outages that matter and improves when you log more trivial tickets, so pair it with a p90 or p95 figure and the incident count.
What does MTTR mean in networking?
The same metric applied to network faults, usually read as mean time to repair or restore: how long a link, device or path stays degraded. Network monitoring tools report it per device or per site, so detection time is included or excluded depending on the tool's configuration.
Related MTTR resources
MTTR is the number SLAs are usually written against, and observability is what shortens the detection and diagnosis part of it. Both definitions sit alongside this one.