Submit

What Is OEE? Meaning, Formula and What the Score Means

OEE meaning explained: Availability x Performance x Quality, what a 'world class' 85% score really means, and why comparing OEE across plants rarely works.

Written & reviewed by Peter Korpak, Founder & Chief AnalystHow we verify
Updated

What is OEE, or overall equipment effectiveness?

OEE multiplies Availability, Performance and Quality into one score for how much of your planned production time made good parts at full speed. Because the three factors multiply rather than add, small losses in each one compound — a plant with three individually solid 90% scores still lands near 73%, not 90%.

Three numbers get multiplied together, and multiplication is the entire point. Add three scores of 90% and you average 90%. Multiply them and you get 73%. OEE is built on multiplication specifically so that a run of small, tolerated losses shows up as one honest number, instead of hiding inside three separate reports that each look fine on their own.

What each factor actually excludes

Availability is Run Time divided by Planned Production Time. It excludes anything the plant scheduled in advance — no shift running, a planned maintenance window, a team briefing — because that time was never meant to make parts. It counts breakdowns, changeovers and material waits as loss, because those minutes were supposed to make parts and didn't.

Performance is actual output measured against the fastest rate the equipment can sustain, not against its rated nameplate maximum. It folds in every small stop and every cycle that ran slower than ideal without separating them into line items, which is why a weak Performance score can have a dozen different, unlogged causes behind it at once.

Quality is good parts divided by total parts made, and it excludes nothing generously: a part that failed inspection and was reworked back to spec is not a good part on the first pass. Scoring it as one is the single most common way OEE gets quietly inflated. The full arithmetic behind all three — a real shift, real downtime minutes, real counts — is worked in the OEE calculation guide.

What "world class 85%" actually means

The 85% figure comes from Seiichi Nakajima's 1984 book on Total Productive Maintenance, where he set it as the level the plants winning Japan's TPM Distinguished Plant Prize were hitting (oee.com). It was never a pass/fail line — most manufacturers today run closer to 60%, and more plants sit below 45% than above 85%.

That gap matters less than what the number can't do: compare plants. Planned Production Time is a local decision, not a physical constant. One plant excludes weekends and books an 8-hour changeover as planned, unavailable time. Another runs continuously and counts that same changeover as an Availability loss. Two lines running identical equipment at identical output can report OEE scores fifteen points apart purely from where each one drew the boundary around "planned." An OEE score is a trend line for one process against itself, not a leaderboard entry against a plant you've never audited.

What the corpus actually shows

Our corpus holds 17 published case studies that name an OEE deployment, inside a wider set of 40 process-optimization deployments reporting some productivity or OEE gain — a median of 25%, ranging from 3% to 80%. That range is the real story. Winstone Wallboards improved OEE by 0.3% and called it a win worth 24 additional production hours a year, on a plant that was evidently already running tight. Zanini Renk moved OEE from 45% to 71% — 26 percentage points — over roughly two years of an Industry 4.0 transformation, because a plant starting at 45% has an entirely different problem to solve.

Neither deployment is more impressive than the other. The size of an OEE gain is mostly a function of how far the starting baseline sat from the ceiling. A vendor quoting "25% average OEE improvement" is quoting a figure that depends entirely on which half of that 3%-to-80% range your plant starts in.

How software changes what you can see

Most of what makes an OEE number hard to trust isn't the arithmetic — it's the data collection underneath it. Agropur found more than 30 hidden hours of annual production time once OEE analytics replaced manual logging, time that had been quietly absorbed into a dozen small, unrecorded stops. Rockline Industries reported a 20%+ OEE increase per cell after adopting FactoryTalk Analytics, largely by putting downtime and slow cycles in one place instead of scattered across paper logs and shift-handoff notes.

Automated data collection doesn't make a machine run faster. It removes the argument about whether the number is even real, which is usually the first fight in any OEE program before a single improvement gets made. The process optimization software hub tracks which platforms show up in these deployments and what they're evidenced for.

Where to go next

  • The arithmetic lives in the OEE calculation guide: a full worked shift, not just the formula.
  • For what a stopped machine actually costs beyond the OEE score, see machine downtime.
  • Every documented process-optimization deployment behind the numbers above sits on the use-case hub, with medians and ranges on the benchmarks page.
  • Fonterra and INX International show the range again from opposite ends: a 20% OEE gain at an already-large dairy cooperative, and a 21.4% gain driven mostly by a 20.5% Performance-factor improvement at a chemical manufacturer.

Related guides