Few indicators, but the right ones
An effective maintenance dashboard fits on one page. Four families of indicators cover the essentials: reliability (is equipment failing?), availability (is it ready when production needs it?), reactivity (what share of work is emergency?) and discipline (is planned work actually executed?).
| Indicator | Calculation | Data required | Main limitation |
|---|---|---|---|
| MTBF (mean time between failures) | Operating hours ÷ number of failures | Dated failure history, running hours | Sensitive to failure logging quality |
| MTTR (mean time to repair) | Total downtime hours ÷ number of repairs | Start and end of each intervention | Mixes repair time and parts waiting |
| Availability | (Required time − downtime) ÷ required time | Production calendar, dated downtime | Depends on the definition of "required time" |
| % emergency work | Emergency hours ÷ total hours | True priority of each order | Priorities are often inflated |
| Schedule compliance | Planned orders completed ÷ planned orders | Frozen weekly schedule | Does not measure execution quality |
Worked example: MTBF and MTTR of a pump
A process pump ran 4,000 hours over six months and experienced 5 failures. Its MTBF is 4,000 ÷ 5 = 800 hours. The five repairs totalled 30 hours of downtime, giving an MTTR of 30 ÷ 5 = 6 hours. Its approximate inherent availability is therefore 800 ÷ (800 + 6) ≈ 99.3%. If MTBF drops to 400 hours the next half with the same MTTR, availability slips to 98.5% — a clear signal to investigate the cause of failures, not just to repair faster.
Pitfalls to avoid
- Multiplying indicators: beyond about ten, nobody looks at them.
- Comparing plants without harmonized definitions (what counts as an "emergency"?).
- Rewarding fast order closure: you get orders closed quickly, not reliable equipment.
- Displaying averages without trend: a KPI with no history or target triggers no action.
A good indicator is not the one that is easy to calculate — it is the one that provokes a decision.