Executive summary
- OEE = availability × performance efficiency × quality rate; the three are coefficients of one another.
- The most common failure is definitional: excluding setup time or minor stops raises the number without any real improvement.
- OEE is a diagnostic, not a target. Managing to the number drives overproduction.
- The maintenance function needs its own indices: MTBF, MTTR, BM rate, PM rate and CM rate.
- Without a loss tree converted into currency, improvement has no priority order.
OEE brings availability, performance efficiency and quality rate together, because the three are coefficients of one another in determining overall effectiveness. All three are the result of losses — so raising OEE is the same thing as reducing loss.
Failure 1: the denominator is wrong
OEE should be calculated against planned production time, defined consistently and agreed with finance. Plants that quietly exclude changeovers, planned maintenance windows or trial runs produce a number that rises without any operational improvement. Reported OEE often falls by 10 to 20 points in the first month of an engagement — not because performance declined, but because the measurement became honest.
Manufacturing example
A steel processing line reported 82% OEE to group. Changeovers and start-up ramps were excluded as non-production activity. Recalculated on planned production time, actual OEE was 61%. The 21-point gap was the improvement opportunity the reporting had hidden for three years.
Failure 2: short stops are invisible
Manual downtime capture reliably misses stops under five minutes, because operators are busy clearing them. On high-speed packaging and assembly lines these micro-stops routinely account for more lost time than every recorded breakdown combined. Automated capture on constraint assets is the highest-return piece of digital instrumentation available.
The maintenance indices that belong beside OEE
OEE measures the outcome at the machine. Measuring the maintenance function itself requires its own indices.
| Index | Meaning | Question it answers |
|---|---|---|
| MTBF | Mean time between failure | Is the equipment becoming more reliable? |
| MTTR | Mean time to repair | Is the maintenance team recovering faster? |
| BM Rate | Breakdown maintenance ratio | How much time is still consumed by reactive work? |
| PM Rate | Preventive maintenance ratio | Is the share of planned work rising? |
| CM Rate | Corrective maintenance ratio | Are we fixing root causes more often? |
| Predictive Rate | Predictive maintenance ratio | How far ahead can we see failure coming? |
Failures 3 to 6
- Treating OEE as a target rather than a diagnostic. When OEE is tied to a bonus, behaviour adapts: longer runs, larger batches, production nobody ordered. Report OEE alongside schedule attainment and finished-goods inventory.
- No loss tree. An OEE percentage cannot be actioned. A loss tree — each loss with a frequency, a duration, an owner and a currency value — can.
- No owner on the floor. A loss assigned to a department is not assigned. It needs a named person, a countermeasure, a due date and a fixed review cadence.
- Monitored but not stabilised. A monitoring platform creates visibility, not stability. If standards, PM and operator capability are unchanged, it will accurately display the same losses month after month.