Eighty-five per cent of what?
The figure is built from three component targets and a manufacturing model most Indian job shops do not match. And without the time base, it carries almost no information.
The figure appears in almost every OEE article ever written: 85% is world class, 60% is typical, 40% is common in plants that have never measured. The numbers are repeated so consistently that they have acquired the status of fact.
They are not wrong so much as incomplete, and the missing part matters enough that quoting 85% without it is close to meaningless. This article sets out where the figure comes from, what it assumes, and why two plants can both be at 85% while one is performing far better than the other.
The 85% figure originates in Total Productive Maintenance literature from Japanese manufacturing in the 1980s, and it is built from three component targets: availability around 90%, performance around 95%, quality around 99%. Multiply those and you get roughly 85%.
That derivation is worth knowing because it tells you what kind of plant it describes: high-volume, low-mix, discrete manufacturing with stable products and long runs. Quality at 99.9% is achievable when you make the same part continuously. Availability at 90% is achievable when changeovers are rare.
A job shop running fifty part numbers a month with changeovers several times a week is not that plant, and never will be. Holding it to a benchmark derived from a different manufacturing model produces either despondency or creative accounting, and frequently both.
OEE is a ratio, and a ratio can be moved as easily by changing its denominator as by improving the numerator. The denominator here is planned production time, and what a plant chooses to exclude from it is a decision, not a fact.
Consider a plant running one shift. If planned production time counts only that shift, and the equipment sits idle for sixteen hours, OEE is unaffected — the idle time was never planned production. The same plant measured against the full day would report roughly a third of the figure. Both calculations are legitimate and they differ by a factor of three.
Now add the usual exclusions: scheduled maintenance, breaks, a shift not run for want of orders, trials and first-off inspection. Each is defensible to exclude, and each removes time that the equipment did not convert into parts. A plant that excludes generously and one that excludes conservatively will report OEE figures that cannot be compared, and — this is the important part — both will believe they can be.
So when a supplier, a competitor or a consultant quotes 85%, the useful question is not whether it is true. It is: 85% of what? Without the time base, the number carries almost no information.
| Situation | What a sensible target looks like |
|---|---|
| You have never measured | No target. Measure honestly for a month first — targets against an unverified number produce reclassification, not improvement. |
| First honest measurement is low | Your own baseline plus a few points, not a published benchmark. The first gains come from losses nobody had recorded. |
| High-mix job shop | Judge the split rather than the product. Changeover frequency caps availability structurally, and no benchmark accounts for your part mix. |
| High-volume, low-mix | The classic benchmarks genuinely apply here, because this is the manufacturing model they were derived from. |
| Comparing plants in a group | One definition and one time base, imposed centrally. Without that, the comparison is arithmetic theatre. |
The only comparison that is reliably meaningful is a plant against its own past, measured the same way throughout. That is unglamorous and it is the truth.
The definition used
ISO 22400-2 or a named variant. A figure without this cannot be checked by anybody.
The time base
What planned production time includes and excludes. This single fact can move OEE by a factor of two or three.
The manufacturing context
Part mix, changeover frequency and shift pattern. An 85% on continuous production and an 85% on a job shop are not the same achievement.
The split
Availability, performance and quality separately. The product alone conceals which of three unrelated problems exists.
MachineWise does not publish benchmark figures for other plants' OEE, and will not until we can publish the methodology alongside them. A number without its definition and time base is not evidence, and the industry has enough of those already.
What we do provide is a measurement you can defend: the definitions are documented in the ISO 22400-2 reference, applied identically across every deployment, and stated on every figure. If you want to compute your own number on both definitions before talking to anyone, the OEE calculator does it in your browser.