Leading KPIs (% planned work, PM compliance, schedule compliance) predict tomorrow’s reliability; lagging KPIs (availability, MTBF, cost) report yesterday’s results. You need both — manage the leading ones to move the lagging ones.
OEE = Availability × Performance × Quality is the single best production-equipment health number; world-class is ~85% (90 × 95 × 99). It multiplies, so one weak factor sinks the whole score.
For the maintenance system itself, watch a short list: equipment availability, schedule and PM compliance, % planned work, backlog weeks, MTBF/MTTR, and cost as % of RAV. More metrics than that usually means less insight.
1 · Leading vs lagging
Every KPI is one of two kinds, and confusing them is why dashboards mislead:
- Lagging indicators measure outcomes that have already happened — availability, MTBF, downtime, total cost. They are the truth, but by the time they move, the cause is months old. You can’t steer by them alone.
- Leading indicators measure the behaviours believed to cause those outcomes — % of work planned, PM/schedule compliance, % proactive vs reactive. They move first, and they are the ones you can actually manage week to week.
The discipline: pick the lagging results you care about, identify the leading behaviours that drive them, and manage the leading ones. If PM compliance and planned work climb but availability doesn’t follow, your theory of what drives reliability is wrong — which is itself valuable to learn.
2 · OEE — the production health number
Overall Equipment Effectiveness rolls three independent losses into one figure, and because they multiply, it is brutally honest — you cannot hide a bad factor behind two good ones.
Each factor points at a different owner: Availability is largely maintenance’s (breakdowns, changeovers); Performance is shared (speed losses, minor stops); Quality is largely process/operations. That is why OEE is a shared operations-and-maintenance number. Move the sliders and watch how a single weak factor drags the whole score down:
Interactive — OEE calculator
Live modelThe OEE waterfall
OEE = A × P × Q. The waterfall starts at 100% and removes each loss in turn — availability (×A), then performance (×P), then quality (×Q) — landing on OEE. “Lost output” is 100% − OEE. World-class reference 85%.3 · The core maintenance KPI set
Beyond OEE, a maintenance organisation needs a short, balanced scorecard. These are the metrics that earn their place — with typical healthy targets (always calibrate to your own context and industry):
| KPI | Formula | Type | Typical target |
|---|---|---|---|
| Availability | uptime ÷ (uptime + downtime) | Lagging | asset-dependent; >95% for critical |
| MTBF | operating time ÷ number of failures | Lagging | rising trend |
| MTTR | total repair time ÷ number of repairs | Lagging | falling trend |
| % Planned work | planned hours ÷ total hours | Leading | >80–85% |
| Schedule compliance | scheduled tasks done ÷ scheduled | Leading | 80–90% |
| PM compliance | PMs done on time ÷ PMs due | Leading | >90% (within window) |
| Backlog | ready backlog hrs ÷ weekly capacity | Leading | 4–6 weeks |
| Maintenance cost as % RAV | annual maint. cost ÷ replacement asset value | Lagging | ~2–3% |
| Reactive ratio | reactive hours ÷ total hours | Lagging | <15–20% |
Availability deserves a note because it links straight to reliability: A = MTBF ÷ (MTBF + MTTR). You raise it two ways — fail less (longer MTBF) or repair faster (shorter MTTR) — and the availability & RAM guide takes that apart in full, including how redundancy multiplies it.
Cost as % of RAV (replacement asset value) is the best single cost benchmark because it normalises for plant size: a refinery and a small process plant can be compared on the same ~2–3% yardstick. Chasing it too low, though, is a classic trap — see below.
4 · The pitfalls
KPIs change behaviour — which is the point, and also the danger:
- Gaming. Any single metric pushed hard gets gamed. Cut maintenance cost % RAV to the bone and reliability collapses a year later; chase MTBF and people stop reporting small failures. Balance leading and lagging, cost and reliability, so no one number can be won at the expense of the others.
- Vanity metrics. “Work orders closed” or “PMs completed” feel productive but say nothing about value. Prefer metrics tied to outcomes (availability, cost, compliance-of-the-right-work).
- Too many. A dashboard with forty KPIs is a dashboard no one acts on. A handful, reviewed in a rhythm, beats a wall of dials.
The KPIs live in the CMMS. Reliable metrics need clean data, and that comes from disciplined work-order close-out and a well-structured asset hierarchy. This is exactly where Bluestream’s CMMS implementation work — increasingly on Microsoft Dynamics 365 — pays off: the system that carries the work also carries the trustworthy numbers to manage it by, and feeds the predictive layer.
Key takeaways
- Separate leading from lagging — manage the leading behaviours (planning, compliance) to move the lagging outcomes (availability, cost).
- OEE = A × P × Q — one honest production number; ~85% is world-class, and one weak factor sinks it.
- Keep a short balanced scorecard — availability, MTBF/MTTR, % planned, schedule & PM compliance, backlog, cost % RAV.
- Beware gaming and vanity metrics — balance the set, tie metrics to outcomes, and keep the list short.