When a water main breaks, a signal goes dark or a lift station alarm sounds, residents judge a public works department by one thing: how quickly service comes back. Mean time to repair puts a number on that speed, but only if the clock, the data and the definitions are consistent across crews. Many agencies calculate it from memory or from spreadsheets that miss half the jobs. This guide explains how to measure mean time to repair from work order data, where the hours actually go, and how a platform like Oxmaint helps crews shorten each stage.
Mean Time to Repair in Public Works
Measure repair speed for streets, water, sewer, signals and facilities from real work order timestamps, then remove the delays that keep assets out of service.
What mean time to repair actually measures
Included in the clock
- Active diagnosis and hands-on repair labor
- Waiting for parts, permits, utility locates or contractors
- Travel time to a remote site
- Testing and verification before return to service
Decisions you must document
- Does the clock start at failure, report or dispatch?
- Does it stop at repair complete or at service restored?
- Are planned shutdowns and scheduled work excluded?
- How are temporary fixes and repeat visits counted?
Related metrics that explain the number
Where the repair hours really go
Detection and reporting delay
Failures found by residents instead of inspections add hours before anyone records a request.
Assignment and travel
Requests sitting in an inbox, unclear ownership or poor location data slow the first response.
Diagnosis
Crews without asset history repeat troubleshooting that a previous visit already completed.
Parts and materials
Missing stock, unknown part numbers and approval steps often become the longest single delay.
Hands-on repair
Actual wrench time is frequently the smallest share, which is why rushing crews rarely helps.
Closeout and verification
Jobs finished in the field but closed days later distort the measured repair time.
Pulling MTTR from work order data
- Asset ID and class
- Request created time
- Assigned and started times
- Completed and closed times
- Work type and priority
- Failure code and cause
- Corrective work only
- Exclude scheduled upgrades
- Separate emergency from routine
- Group by asset class or district
- Remove duplicates and cancelled jobs
- Are timestamps entered live or later?
- Do closed dates match field completion?
- Are outliers explained by a note?
- Is the sample large enough?
Averages hide the worst outages
- Many quick fixes pull the number down
- One week-long outage disappears into the mean
- Emergency and routine jobs are blended
- Report median alongside the mean
- Track the slowest ten percent of repairs separately
- Segment by priority and asset class
See repair time by asset class without building a spreadsheet
Review how work order timestamps turn into dashboards crews and managers can both trust.
Practical ways to shorten repair time
| Delay | Practical fix | How software helps |
|---|---|---|
| Slow assignment | Define dispatch rules and response targets by priority | Work orders route to a crew with status and ownership visible |
| Repeated diagnosis | Record failure cause and fix on every job | Asset history appears on the mobile work order |
| Parts shortages | Stock critical spares and set reorder points | Inventory is linked to assets and flags low stock |
| Repeat failures | Review the same asset after the second repair | Reports surface assets with frequent corrective work |
| Late closeout | Close jobs from the field at completion | Mobile updates capture accurate finish times |
| Unplanned breakdowns | Shift routine inspections into planned work | Preventive schedules and inspections generate tasks automatically |
Repair priorities for typical public works assets
A repair improvement cycle crews can follow
Mean time to repair FAQs
Should MTTR start at the report or at dispatch?
Pick one definition and keep it. Starting at the report reflects what residents experience, while dispatch isolates crew performance.
Is a lower MTTR always better?
Not if rushed temporary fixes cause repeat failures. Pair it with repeat repair counts and quality checks.
How many repairs do we need for a useful number?
Enough that one outlier does not swing the average, so group related assets. Report the median alongside it.
Can software calculate this automatically?
Yes, when work orders carry reliable timestamps. You can book a demo to see the reports.
Where should a small department begin?
Track one asset class for a month, then sign up to log every repair with start and finish times.
Restore service faster and prove it with data
Give crews one place for requests, parts and repair history, and give managers a number they can defend.
