OEE is the most quoted and most misused number in manufacturing. How to calculate it honestly, why 85% is the wrong target, where the hidden capacity usually is, and how to turn the number into decisions.
A plant manager once showed us a dashboard proudly reporting 82% OEE on his bottleneck line. Two days later, after timing the line ourselves, the real figure was 54%. Nobody had lied. The ideal cycle time in the system was the one the operators considered "normal", micro-stops under five minutes were never logged, and changeovers were booked as planned downtime. The line looked world class on paper and was quietly losing a third of its capacity.
That gap between the reported number and reality is why OEE, Overall Equipment Effectiveness, is both the most useful and the most abused metric in manufacturing.
The calculation, done honestly
OEE multiplies three ratios:
- Availability: run time divided by planned production time. Breakdowns, changeovers, waiting for material and waiting for an operator all count as losses.
- Performance: actual output against what the machine could have made at its ideal cycle time while it was running. Slow running and micro-stops live here.
- Quality: good parts first time divided by total parts made. Scrap and rework both count against it, including start-up scrap.
A line that is available 80% of the time, runs at 78% of ideal speed and makes 97% good parts has an OEE of 0.80 × 0.78 × 0.97 = 60.5%. Each factor looks respectable on its own. Multiplied together, four hours of every ten have gone somewhere.
The three ways OEE gets inflated
- The ideal cycle time is set too slow. It should be the best demonstrated or design rate of the equipment, not the standard the costing system uses. Set it 15% too slow and Performance is flattered by 15% forever.
- Changeovers are hidden as planned downtime. If you plan to change over, it is still lost capacity. Keep it inside Availability, or you will never see the case for a changeover kaizen.
- Micro-stops are not recorded. A 40-second jam, twenty times a shift, is over thirteen minutes a shift that no operator will ever write down. Only automatic data capture or a stopwatch study will show it.
Why 85% is the wrong target
The famous "world class OEE is 85%" figure comes from discrete, high-volume, dedicated production. It is a useful headline, but as a target it does damage. A high-mix moulding cell with eight changeovers a day will never reach it, and chasing it encourages exactly the inflation described above. A continuous process line should be well above it.
The only fair comparison for OEE is the same asset against itself over time. Use it to find losses and track improvement, not to rank plants or punish shifts.
Where the hidden capacity usually is
When we break down OEE losses on a typical first study, the pattern is remarkably consistent:
- Speed losses and micro-stops are larger than anyone expects, often a third of all losses, because they are invisible without measurement.
- Changeovers are the biggest single Availability loss on high-mix lines, and the most fixable: halving them is common.
- Breakdowns get all the management attention but are often smaller than the first two combined.
- Quality losses look small in percentage terms but cost the most per unit, because every scrapped part has already consumed machine time, material and labour.
The most powerful question you can ask of any OEE report is: "Which of the six big losses is biggest on this asset this month?" Breakdowns, set-up and adjustment, small stops, reduced speed, start-up rejects and production rejects. If the answer is not obvious within ten seconds of looking at your report, it is not a management tool yet.
What one point of OEE is worth
Put a value on it. Take a bottleneck line that runs two shifts, five days a week, with contribution margin of $900 per hour of output. Each additional point of OEE is roughly 40 extra productive hours a year, so about $36,000 of margin. Moving from 55% to 65% is worth over a third of a million dollars a year, without buying a single machine.
That is the conversation that gets OEE funded: not "our OEE is low", but "the hidden capacity on this line is worth this much, and here is how we will release it".
Turning the number into action
- Measure honestly first. Fix the ideal cycle time, include changeovers, and capture micro-stops for at least two to four weeks before you set any targets.
- Focus on the constraint. OEE improvement on a non-bottleneck machine releases capacity you cannot sell.
- Attack the biggest loss with the right tool: SMED for changeovers, autonomous maintenance for small stops, and DMAIC for chronic quality losses. We described the method in DMAIC in plain English.
- Make it visible at the line. A shift-by-shift chart of the top three losses, discussed for ten minutes every morning, does more than any monthly report. Visual management only lasts if the workplace supports it, which is why 5S that lasts matters more than it sounds.
A simple test for your own OEE
Pick your most important machine. Stand next to it for one hour with a stopwatch and a notepad, and write down every stop, however short, and the actual cycle time. Calculate OEE for that hour and compare it with your system's figure for the same hour.
If the two numbers are within a few points of each other, you have a measurement system you can trust. If they are not, the gap between them is your first improvement project, and very often the most valuable one.
