Which KPIs show how well a tracker or MDVR installation holds up?
A device installation is judged on two things: how often the hardware fails, and how fast a failure is put right. These eight KPIs cover both, for trackers and for MDVRs.
| KPI | What it shows | Built from |
|---|---|---|
| Annualized failure rate | How often installed hardware fails, scaled to a year so a month can be compared with a year. | Faults, units in service, days |
| Dead-on-arrival rate | Units that fail at the install test or soon after. Points at stock handling and incoming checks. | Faulty early, units installed |
| Install rework rate | Re-visits caused by workmanship rather than the device. | Install re-visits, units installed |
| First-time fix rate | Jobs closed in one visit. Low values mean missing spares or a wrong diagnosis. | First-visit fixes, jobs closed |
| Mean time to repair | Average days from the fault report to a working unit. | Repair days, jobs closed |
| Service availability | Share of device-days in which units could record and report. | Down days, units, days |
| Storage failure rate | SD cards and disks that fail in MDVRs, scaled to a year. | Storage failures, MDVRs, days |
| Clip retrieval success | Footage requests where the clip was really on the device. | Clips found, clips requested |
Where do the counts come from?
Use records you already keep. Each count should point at a document someone can open.
- Device faults and storage failures: the swap and RMA custody forms, one per unit removed.
- Dead on arrival and install re-visits: the install test sheet and the installer's job list.
- Service jobs, first-visit fixes and repair days: the job sheets with report and close dates.
- Device-days out of service: the days each faulty unit stayed silent, from the platform export or the job sheet.
- Clip requests: the footage request log with the outcome of each request.
Worked example
- 240 units in service for 30 days with 1 hardware fault: 1 / 240 x 365 / 30 = 5.07% a year. Against a 5% target with a 2-point band, that is near.
- 20 units installed, none dead on arrival, 1 install re-visit: DOA 0%, rework 5%. Rework misses a 3% target.
- 10 jobs closed, 8 on the first visit, 24 repair days: first-time fix 80%, MTTR 2.4 days.
- 30 device-days down out of 240 x 30 = 7,200: availability 99.58%.
- 1 storage failure in 120 MDVRs over 30 days: 10.14% a year. 19 of 20 clips found: 95%.
How we calculate
AFR (%/yr) = faults / units in service x 365 / days x 100
MTBF (h) = units x days x 24 / faults
DOA (%) = faulty early / units installed x 100
Rework (%) = install re-visits / units installed x 100
First-time fix (%) = first-visit fixes / jobs closed x 100
MTTR (days) = repair days / jobs closed
Availability (%) = (1 - down days / (units x days)) x 100
Storage failure (%/yr) = storage failures / MDVRs x 365 / days x 100
Clip retrieval (%) = clips found / clips requested x 100Met: at or better than the target. Near: inside the band. Missed: outside the band. A KPI with no base, such as no installs, is not scored.
How many spare units should you keep?
Spares cut repair time more than any other step. The tool takes the measured failure rate, works out faults a year, and keeps enough units to cover the lead time of a repair or a new delivery.
Faults a year = AFR / 100 x units in service
Spares = round up (faults a year x lead time days / 365)
12.17 faults a year x 14 / 365 = 0.47 → 1 spare unit
Is this the same as a device health report?
No. A health report reads a device export and lists which units are silent today. This scorecard works on a month of service records and tells you how the installed base and its support performed over time. Use the health report to find the units, and this page to judge the trend.
Assumptions and limits
- Targets and bands are examples. Set them from your service contract or your own history.
- Small fleets swing a lot: one fault in 50 units over a month is already about 24% a year. Judge small fleets over a quarter or a year.
- Faults are counted per unit, not per visit. Wiring, SIM and configuration issues are not device faults.
- Availability counts the days you enter. It does not read any device data.
- Everything is calculated in your browser. If you ask for the emailed report, your inputs are sent to our server with the request and kept for up to 45 days.