Maintenance KPIs are the small set of numbers that tell you whether maintenance is controlling equipment reliability and cost, or being controlled by breakdowns. The useful set splits into leading indicators (work discipline you can change this week: compliance, backlog, planned-work share) and lagging indicators (outcomes that follow months later: MTBF, downtime, cost).

Most plants track either too few, one downtime number nobody trusts, or too many, a 40-metric report nobody reads. Ten is enough. This guide gives the ten with formulas, the leading-versus-lagging logic that connects them, honest notes on targets, and a one-screen dashboard layout.

Why do leading and lagging indicators both matter?

Lagging indicators measure results: how often equipment failed, how long repairs took, what maintenance cost. They are the truth, but they arrive too late to manage with, this quarter's MTBF was determined by decisions made months ago. Leading indicators measure the behaviors that produce those results: did the PMs happen, did the schedule hold, is the backlog controlled? They are actionable weekly, and they predict where the lagging numbers will go.

The management rule that follows: review leading indicators weekly and act on them; review lagging indicators monthly or quarterly and use them to check whether the leading discipline is working. A plant holding 90%+ PM compliance and 80%+ schedule compliance for two quarters will see MTBF rise and emergency work fall. If it does not, the PM content is wrong, which is itself a finding.

The KPI tree: leading discipline drives lagging outcomesLEADING · weeklybehaviors you controlPM complianceschedule complianceplanned work %ready backlog (wks)wrench timedriveLAGGING · monthlyoutcomes that followMTBF ↑MTTR ↓unplanned downtime ↓emergency work % ↓maintenance cost ↓show up asPLANT OUTCOMESavailability · OEEcost per unitmanage the left column weekly; the middle and right follow with a lag of months
The KPI tree. Weekly discipline metrics on the left drive the reliability outcomes in the middle, which surface in plant-level availability and cost.

The 10 maintenance KPIs

  1. PM compliance (leading). Scheduled PMs completed within their due window. The heartbeat of the PM schedule; the commonly cited target is 90%+.
  2. Schedule compliance (leading). Scheduled work-order hours completed in the week they were scheduled. Measures whether the weekly plan survives contact with reality, the core scorecard of planning and scheduling.
  3. Planned work percentage (leading). Share of executed work hours that were planned before execution. High planned share is the single strongest signature of a proactive department.
  4. Emergency work percentage (lagging). Share of work hours on drop-everything jobs. The firefighting gauge; it falls as PM and planning discipline rise.
  5. Ready backlog in crew-weeks (leading). Planned, parts-in-hand work divided by weekly crew capacity; commonly cited healthy range 2–4 crew-weeks. Full treatment in our backlog guide.
  6. Wrench time (leading). Share of a technician's day spent on tools rather than hunting parts, permits, or information. Honest note: measuring it well requires observation sampling, and typical unmeasured plants are far lower than they assume. Even a rough quarterly sample exposes the waste.
  7. MTBF (lagging). Operating time ÷ failures, per critical asset. Rising MTBF is the point of all the discipline above. Formula and misuses in the MTBF guide.
  8. MTTR (lagging). Failure downtime ÷ repair count, with its four components (detect, diagnose, wait, repair) tracked separately where possible, see the MTTR guide.
  9. Unplanned downtime (lagging). Hours per period per line, from machine data where possible rather than manual logs. Feeds availability and OEE.
  10. Maintenance cost (lagging). Total maintenance spend, best normalized per unit produced (or, in asset-heavy operations, as a percent of replacement asset value). Watch the mix shift from reactive spend to planned spend, not just the total.
#KPIFormulaTypeCadence
1PM complianceon-time PMs ÷ scheduled PMs × 100LeadingWeekly
2Schedule compliancecompleted scheduled hours ÷ scheduled hours × 100LeadingWeekly
3Planned work %planned hours executed ÷ total hours executed × 100LeadingWeekly
4Emergency work %emergency hours ÷ total hours × 100LaggingWeekly
5Ready backlogready backlog hours ÷ crew capacity hours/weekLeadingWeekly
6Wrench timeon-tool time ÷ paid time × 100 (sampled)LeadingQuarterly
7MTBFoperating time ÷ number of failuresLaggingMonthly
8MTTRfailure downtime ÷ number of repairsLaggingMonthly
9Unplanned downtimeunplanned down hours per line per periodLaggingWeekly
10Maintenance costspend ÷ units produced (or % of asset value)LaggingMonthly

What should the dashboard look like?

One screen, three bands: leading discipline on top (because it is reviewed weekly and acted on), reliability outcomes in the middle, money at the bottom. Trends beat snapshots, every tile should show direction, not just the current value.

One-screen maintenance dashboard layoutMaintenance, week of …leading · weekly reviewPM COMPLIANCE91% ▲SCHED COMPLIANCE78% ▲PLANNED WORK %64% ▲READY BACKLOG3.2 wkEMERGENCY %18% ▼lagging · monthly trendsMTBF BY CRITICAL ASSETMTTR + WAIT COMPONENTUNPLANNED DOWNTIME HRSMAINT COST / UNIT · reactive vs planned mixTOP 5 OFFENDERS · failures × downtime cost1 filler #2 · 2 case packer · 3 compressor …
A one-screen layout: leading tiles on top with week-over-week direction, reliability trends in the middle, cost and a top-offenders list at the bottom.

The build order matters more than the layout: start with the four metrics you can compute from data you already trust (usually PM compliance, backlog, downtime, cost) and add the rest as data quality allows. A beautiful dashboard on bad data teaches people to ignore dashboards. The classic failure mode is that work orders, downtime logs, and cost data live in three systems that do not talk, which is a data-unification problem before it is a KPI problem. Connecting those sources into one layer with live dashboards is the job described on our platform overview; the CLS case study shows automated daily reporting replacing hand-built spreadsheets.

Where do the benchmarks honestly come from?

Ten numbers, one screen, reviewed on a rhythm, leading weekly, lagging monthly. That cadence, more than any individual metric, is what separates plants that manage reliability from plants that read about it. For how the whole system fits together, start at the equipment reliability hub.