Operations → Reliability
MTTR
The elapsed time required to restore acceptable service after a production failure or degradation.
What it measures
Mean Time to Restore (MTTR) measures the time from a qualifying service disruption until acceptable service is restored. Restoration may come from rollback, failover, traffic shifting, feature-flag disablement, capacity changes, or a forward fix.
Because incident durations are often skewed, teams should inspect median and percentile restoration times as well as the arithmetic mean. The measurement boundary must distinguish restoration from complete root-cause remediation, which may occur later.