Key points
- Track a few KPIs well rather than many badly. Three to five, reviewed every month, beat a dashboard of twenty nobody reads.
- Pair a result KPI with a cause KPI. Availability tells you something is wrong; PM compliance and repeat failures tell you why.
- Your own trend is the benchmark that matters. Published targets assume a site, an asset mix and definitions that are not yours.
- Every KPI is only as good as the records behind it. A breakdown nobody logged is a failure the numbers will never show.
The ten KPIs at a glance
| KPI | Question it answers | Better when |
|---|---|---|
| Availability | Are our assets there when we need them? | Higher |
| MTBF | How long does an asset run between failures? | Higher |
| MTTR | How long does a repair take? | Lower |
| PM compliance | Is planned maintenance done when it is due? | Higher |
| Planned maintenance percentage | How much of our time is planned rather than firefighting? | Higher |
| Backlog | How much work is waiting, and for how long? | Stable and young |
| Work order completion time | How long from raising a job to closing it? | Lower |
| Repeat failures | Are we fixing causes or symptoms? | Lower |
| Maintenance cost per asset | Which assets cost the most to keep running? | Lower, in context |
| First-time fix rate | Are jobs fixed on the first visit? | Higher |
1. Availability
Availability = uptime ÷ (uptime + downtime) × 100
The share of time an asset was able to work. It is the KPI management asks about first, because it is the one they feel. Decide up front which hours count: calendar hours (every hour in the period) or scheduled hours (only the shifts you planned to run). Calendar hours are simpler and cannot be argued with; scheduled hours are fairer to a site that works one shift. Either is fine, as long as you never compare one with the other.
Availability, MTBF and MTTR are worked through in detail, with an example, in MTBF, MTTR and availability.
2. MTBF (mean time between failures)
MTBF = operating time ÷ number of failures
How long an asset runs, on average, before it breaks down. Use operating hours from the hour meter where you have one; calendar time flatters an asset that sat idle. MTBF is most useful per asset or per asset type, where it shows which machines fail most often. Across a whole site it mixes machines that share nothing and says little.
3. MTTR (mean time to repair)
MTTR = total repair time ÷ number of repairs
How long a repair takes once a fault is reported. The useful part is splitting it: time waiting for someone to respond, time diagnosing, time waiting for parts and time actually working. A high MTTR caused by waiting for parts is a stores problem, not a skills problem, and only the split tells you which. A few long repairs can drag the average up, so look at the median as well.
4. PM compliance
PM compliance = preventive services done on time ÷ preventive services due × 100
The share of planned services done when they were due. It is the cause behind most availability problems, so it is the KPI to watch most closely. The trap is defining "on time". Write the tolerance down, for example within 10% of the interval for an hours-based service or within a week for a monthly one, and apply it the same way every month. A compliance figure whose definition changes is a number about nothing.
5. Planned maintenance percentage
Planned maintenance % = hours on planned work ÷ total maintenance hours × 100
How much of your team's time goes to work you chose to do, compared with work that chose you. It is the clearest single sign of whether a preventive maintenance programme is paying off: as it works, hours move from repairs to planned services. You need work orders typed consistently as preventive or reactive for this to mean anything, which is why a short list of work order types matters. If you do not record hours, count work orders instead and say that is what you counted.
6. Backlog
Backlog (weeks) = estimated hours of open work ÷ available maintenance hours per week
How much approved work is waiting. Expressed in weeks of crew time it becomes comparable between teams of different sizes. Without hour estimates, track the number of open work orders and the age of the oldest ones. A growing backlog means demand is outrunning capacity; a backlog that is steady but full of old jobs means some work is being quietly abandoned. Both deserve a conversation.
7. Work order completion time
Completion time = date closed − date raised, averaged over closed work orders
How long jobs stay open. Split it by priority, because an average that mixes critical repairs with low-priority odd jobs hides both. A critical job taking days is a problem; a low-priority job taking two weeks may be exactly the plan. If jobs are often put on hold, record why: waiting for parts, a permit, a contractor or access. The reasons are where the time goes.
8. Repeat failures
Repeat failure rate = failures on the same asset with the same cause within N days ÷ all failures × 100
Faults that come back. A repeat within, say, 30 days usually means the symptom was fixed and the cause was not. This KPI needs failures classified by cause, consistently, so it pays to use a fixed list of breakdown codes rather than free text. The short list of assets with the most repeats is often the best-value improvement work a maintenance team can do.
9. Maintenance cost per asset
Cost per asset = parts + labour + outside services charged to the asset over the period
What each asset costs to keep running. For machines with meters, dividing by operating hours gives cost per hour, which makes machines of different ages and workloads comparable. Over time this is the honest basis for a repair or replace decision. It only works if parts and labour are recorded against the asset: a part issued to "the workshop" is a cost that belongs to no machine.
10. First-time fix rate
First-time fix rate = jobs completed on the first visit ÷ all jobs completed × 100
The share of jobs finished without a second trip. Low figures usually point to work orders that did not describe the fault well, technicians arriving without the right part, or skills gaps. A photo of the fault on the work order and a parts list for common jobs are the cheapest ways to move it.
Which KPIs to start with
If you are starting from nothing, do not try to measure all ten. Start with what your records can already support, and add the next one when the first is trusted:
- Backlog and its age. Needs only open work orders with dates. It is the quickest way to see whether work is under control.
- PM compliance. Needs a preventive plan per asset and completed services recorded. It shows whether the plan is real.
- Availability and MTTR. Need breakdowns logged with a start and an end. They show what the plan is buying.
- Planned maintenance percentage and cost per asset. Need work orders typed consistently and parts recorded against assets. These come once the basics are habit.
Mistakes that make maintenance KPIs lie
- Unlogged breakdowns. A fault fixed over the phone and never recorded improves MTBF and availability while the machine gets worse.
- Mistakes marked as repairs. A breakdown logged in error should be cancelled, not closed as fixed, or a failure that never happened enters every figure.
- Changing definitions. Moving from calendar to scheduled hours, or widening the on-time tolerance, makes this month incomparable with last month.
- Averages over everything. One site-wide MTBF across generators, forklifts and pumps says little. Split by asset type.
- Targets copied from elsewhere. A figure from another industry or another company's report assumes their definitions. Set targets from your own last six to twelve months.
- KPIs used to blame. If a number is used to punish, people stop logging what makes it worse. Use KPIs to find problems, not culprits.
Maintenance KPIs in Solmex
Solmex calculates these figures from the records your team keeps day to day, so there is no separate spreadsheet to fill in. Breakdown analysis gives availability, MTBF, MTTR split into response and repair phases, repeat failures and the assets and causes behind most downtime. The preventive maintenance dashboard shows services due, overdue and done on time, and lists assets with no plan rather than leaving them out. Work orders carry their type, priority and dates, and the cost ledger collects fuel and parts against each asset. Where a figure cannot be calculated honestly, for example cost with parts not yet priced, Solmex says so instead of showing a number. See reports and dashboards for details.