Mean Recovery Time (MTTR)

Definition

Mean Recovery Time (MTTR) is the arithmetic mean, in minutes, of the elapsed time from incident detection to verified resolution, computed over all dispatch tickets that reached a RESOLVED state within the trailing window. The clock starts at the detection timestamp (the metric breach or fault event, not ticket creation) and stops at operator-verified resolution, not at the provisional auto-clear. Cancelled and duplicate tickets are excluded.

Why it matters

MTTR is the truest measure of how quickly the facility returns from an off-nominal state to health. Queue depth tells you how much is open; MTTR tells you how fast it gets fixed. A rising MTTR with stable queue depth means individual incidents are getting harder — parts shortages, skill gaps, or recurrent root causes — while a low, stable MTTR underpins every recovery-time service commitment the Directorate holds. CHLORA publishes it on the Operations Dashboard and uses it to tune crew dispatch priority.

Formula

MTTR is the mean of per-ticket recovery durations over the trailing window:

MTTR = ( 1 / N ) · Σ ( t_resolved,i − t_detected,i )
        over tickets i resolved in trailing 60 min, in minutes

where:
  t_detected  = timestamp of the originating breach/fault event
  t_resolved  = timestamp of operator-verified resolution
  N           = count of qualifying resolved tickets in window

Exclusions:
  - CANCELLED and de-duplicated child tickets
  - tickets resolved < 30 s (auto-clear noise)

Segmentation (published alongside, not in headline):
  MTTR_P1, MTTR_P2, MTTR_P3, MTTR_P4  by severity band
  P95 recovery time reported to expose long-tail incidents.

If N < 5 in window, MTTR carries a LOW_SAMPLE confidence flag.

Inputs

Units & Scale

MTTR is reported in minutes to whole-minute precision with a P95 long-tail figure alongside. It is a window mean, not additive across zones; a facility MTTR weighted by ticket count is published rather than a simple zone average. The trailing window is 60 minutes, re-evaluated hourly. A LOW_SAMPLE confidence flag is attached when fewer than five qualifying tickets fall in the window, since small N makes the mean volatile.

Sampling & Source

• Recomputed every 1 hour by CHLORA over a trailing 60-minute window of resolved tickets.
• Detection and resolution timestamps both sourced from the CHLORA dispatch ledger.
• Recovery clock runs from breach detection to operator-verified resolution, not ticket creation to auto-clear.
• Stale / low-sample handling: window with N < 5 tickets → value held with LOW_SAMPLE; no resolutions → last value carried forward, marked STALE.

Thresholds

Recent Trend

Facility MTTR in minutes, last 14 hourly windows:

Interpretation Guidance

Related Metrics

Related SOPs

• SOP Library — incident response and root-cause-review procedures.
• Window root-cause review on WARN/CRIT MTTR — see the SOP Library.
• Recovery-time monitoring & long-tail escalation — Monitoring Systems.