What Is OEE and How Do You Calculate It?

In every plant the same sentence comes up: the machine runs all the time. Measurement usually tells a different story. A shift is eight hours, but the machine may have genuinely produced for five and a half of them; the rest went to changeovers, waiting, small stops and parts that ended up as scrap.

OEE — overall equipment effectiveness — compresses that picture into one number. Its purpose is not to audit anyone. It is to show where the loss sits, because three very different causes produce the same shortfall and each needs a different fix.

OEE is the product of three ratios

Component	What it measures	Formula
Availability	Did the machine run when it was supposed to	Run time ÷ planned production time
Performance	Did it run at its own rated speed	(Ideal cycle time × total count) ÷ run time
Quality	How many parts came out good	Good count ÷ total count

All three are percentages, and OEE = availability × performance × quality. Because it is a product, one weak component drags the result down quickly: three ratios of 90% each give an OEE of 72.9%.

A worked shift

Take an eight-hour shift. Of the 480 minutes, 30 are a scheduled break, so planned production time is 450 minutes. Breakdowns and changeovers took 60 minutes, leaving 390 minutes of run time. The machine's ideal cycle time is 30 seconds per part, or two parts a minute. The counter shows 620 parts at the end of the shift, and quality control rejected 20 of them.

Component	Calculation	Result
Availability	390 ÷ 450	86.7%
Performance	(0.5 min × 620) ÷ 390	79.5%
Quality	600 ÷ 620	96.8%
OEE	0.867 × 0.795 × 0.968	66.7%

The biggest loss here is performance: the machine is running, but below its own rated speed. Small stops, material waiting and slow starts accumulate in that component, and they usually go unnoticed — from the shop floor the machine simply looks busy.

The widely quoted target is 85%, which means roughly 90% availability, 95% performance and 99% quality. In practice most sites land between 50% and 70% on their first measurement. A low first number is not a bad sign; it is the first time the loss has been visible at all.

Five common mistakes

• Counting planned downtime as loss. Breaks, planned maintenance and unstaffed hours come out of planned production time. Record them as loss and OEE stays permanently low, at which point nobody trusts the number.
• Deriving ideal cycle time from an average. The ideal is the rated or best-observed speed. Use last month's average and performance will always read close to 100%, which makes the measurement pointless.
• Ignoring scrap and rework. A part rescued by rework is still a quality loss; it was not made right the first time.
• Looking at daily averages. Losses vary by shift and by product. An average hides the bad shift inside the good one.
• Using OEE as a staff scorecard. The measurement is about the machine. The day the number becomes a personal rating, the data degrades: stops go unrecorded and reasons stop reflecting reality.

Measuring without touching the machine's program

The common worry is justified: modifying a running line risks stopping production. The good news is that OEE does not require it. There are three routes, and they combine well.

• Read the machine's own counter. Most lines already count parts and downtime, and the operator panel shows it. Taking the data from there carries the least risk.
• Add an external counter or sensor. A counter wired to a signal output, or a sensor mounted on the line, records count and stoppage independently of the machine. Nothing in the machine's software changes.
• Operator entry. Counts and stop reasons entered on a tablet at the end of the shift are enough for the first month. Only a person can supply the reason for a stop; a sensor knows the duration, not the cause.

What all three require is that the data leaves the machine. The numbers have to land somewhere and turn into a report someone reads weekly, otherwise the measurement quietly dies within a month.

What to do in the first week

• Start with one machine — the bottleneck, or the one that stops most often.
• Keep stop reasons down to ten. A thirty-item list never gets filled in; a three-item list says nothing.
• Record per shift. Daily totals mislead at this stage.
• Treat the first fortnight as a baseline photograph rather than a target.
• Hold a fifteen-minute review once a week, pick the single largest loss and work on that one.

How Yamanlar Bilişim helps

Your own team can set this up. If you would rather bring someone in, our part is this: take the data from the existing panel or an external counter, collect it in one place, and turn it into a per-shift report people can actually read. We do not modify the machine's program, and installation happens during planned downtime. The rest of our work for production sites is on the manufacturing and workshops page.

Frequently asked questions