MTTR / MTTD
The averages for how long incidents go unnoticed and how long service takes to recover, used to find where response is slow.
See it
What it is
MTTD is mean time to detect: the average gap between an incident starting and the team or system recognizing it. MTTR here means mean time to recover: the average gap between the start and restored service. Together they separate slow discovery from slow recovery. Google's DORA research tracks a close cousin, time to restore service, as one of its four delivery metrics, which is usually why leadership asks for this number.
Reach for them when improving incident response across many events, not when judging one outage. Record start, detection, mitigation, and restoration timestamps consistently, then compare trends by service and incident class.
Gotcha: MTTR is also expanded as repair, resolve, or respond, and each version uses different endpoints. Publish the exact clock definition beside the chart. Means also hide a few terrible incidents, so show the count, median, and 90th percentile with them.
Ask AI for it
Add MTTD and MTTR reporting to this incident system. Store started_at, detected_at, mitigated_at, and restored_at as PostgreSQL timestamptz columns with ordering constraints. Define MTTD as detected_at minus started_at and recovery time as restored_at minus started_at. Build a rolling 90-day Grafana dashboard with count, average, median, and p90 by service and severity, using PostgreSQL percentile_cont for percentiles. Keep drills in a separately labeled series, flag missing timestamps, and display the formulas beside every chart.