What is MTTR (mean time to repair or recover)?
MTTR is calculated by adding up the time spent on a set of incidents and dividing by the number of incidents. The difficulty is the R. In common use it stands for repair or recovery, and some organizations also use it for respond or resolve. Atlassian, for example, notes that the label can refer to four related but different measurements.
Mean time to repair usually measures the time to fix the faulty component, from when work begins until it is working again. Mean time to recover or restore usually measures the whole outage as the customer sees it, from the start of the failure until service is back. The second number is longer, because it includes detection, response, travel and testing.
Because there is no single definition, two MTTR figures from different providers cannot be compared unless the start and stop points are the same. MTTR is an average, so it can also hide a few very long incidents. It is a reporting measure and not a promise, which is what an SLA target is.
How it shows up in practice
- Write down the definition in the contract: which event starts the clock, which stops it, and which hours count.
- Separate on-site arrival time from fix time, because travel and access can dominate a field service incident.
- Ask for the distribution of times, not just the average, so a few very long incidents are visible.
- Compare like with like. Do not compare a repair MTTR from one provider with a recovery MTTR from another.