Mean Time to Repair (MTTR) explained
Mean Time to Repair measures how long your jobs take — specifically, the time from when a work order is dispatched to when it's resolved. It's one of the clearest signals of field efficiency. On the Field Service overview it's a line chart of the monthly average, with a set of summary statistics.
Why median, not just the average
The word "mean" is in the name, but a plain average is easy to mislead. If nine jobs take an hour and one drags on for two days, the average jumps well above what a typical job actually takes. So Roldesk leads with the median — the middle job, where half finish faster and half slower. The median is the honest answer to "how long does a normal ticket take?"
We still show the mean right beside it. Comparing the two tells you something useful:
- Median ≈ mean → your jobs are consistent.
- Mean much higher than median → a few long outliers are dragging the average up. Worth investigating what those jobs had in common.
P90 and P95 — the planning numbers
Below the chart you'll see P90 and P95 once there's enough data:
- P90 = 90% of jobs finish within this time.
- P95 = 95% of jobs finish within this time.
These are the numbers to use when you promise a timeframe to a customer or set an SLA. Promising "we'll be done in the median time" fails half the time; promising the P90 keeps you honest for 9 out of 10 jobs. Percentiles only appear once you have a reasonable sample (a handful of finished jobs) — below that they'd be too noisy to trust, so Roldesk hides them until they're meaningful.
The reference lines on the chart
The median and mean are drawn as horizontal lines across the chart (median solid, mean dashed), so each month's dot reads against them at a glance — you can instantly see which months ran hotter or cooler than typical.
How to act on it
- MTTR creeping up? Look at travel time, parts availability, and whether the right-skilled tech is being assigned.
- Big gap between median and P90/P95? A minority of jobs are running very long — often a parts or diagnosis bottleneck.
- Want a lower MTTR? Better first-visit parts stocking and dispatching the nearest qualified tech are the usual levers.
Next: Measuring service quality: SLA & callback rate · Understanding your Field Service metrics