What Is OEE? How to Calculate It, With a Worked Example

OEE — Overall Equipment Effectiveness — is a single percentage that answers a blunt question: of the time a machine was supposed to be making good parts, what fraction actually produced good parts at full speed? It multiplies three factors — Availability, Performance, and Quality — so every common form of loss (downtime, slow cycles, scrap) drags the number down.

The math takes five minutes per shift with numbers you probably already write down. This guide defines each factor precisely, walks a full example, and covers the traps that make OEE figures meaningless if you're not careful.

The three factors, precisely

Availability = Run Time / Planned Production Time. Planned production time is the shift minus planned exclusions (scheduled breaks, planned maintenance, no-demand time — whatever you decide doesn't count against the machine). Run time is what's left after unplanned stops: breakdowns, changeovers, waiting on material, waiting on an operator.
Performance = (Ideal Cycle Time x Total Count) / Run Time. It asks: while the machine was running, how close to its ideal speed did it run? Slow cycles and brief micro-stops that never get logged as downtime show up here.
Quality = Good Count / Total Count. Only parts that pass first time count as good — reworked parts are losses in a strict OEE calculation, because the machine time to make them right the first time was consumed.

OEE = Availability x Performance x Quality. Because it's a product, mediocrity compounds: a machine at 90% on each factor is not at 90% OEE — it's at 72.9%. That compounding is the whole point. It stops you from celebrating uptime while quietly scrapping a tenth of what you make.

Tip Ideal cycle time means the fastest the machine can reliably run — the nameplate or best demonstrated rate — not the rate you currently plan for. If you use a padded 'standard' rate, Performance flatters itself and losses disappear from view.

Worked example: one machine, one shift

Take an 8-hour (480-minute) shift on a CNC lathe making one part with an ideal cycle time of 30 seconds (0.5 minutes) per part.

Planned production time: 480 minutes minus a 30-minute scheduled lunch = 450 minutes.
Unplanned stops during the shift: one 27-minute breakdown, an 18-minute changeover, and 15 minutes waiting on material = 60 minutes. Run time = 450 - 60 = 390 minutes.
Availability = 390 / 450 = 86.7%.
Total parts produced: 660. At the ideal 0.5 min/part, 660 parts should have taken 330 minutes. Performance = 330 / 390 = 84.6%.
Good parts first time: 640 (14 scrapped, 6 sent to rework). Quality = 640 / 660 = 97.0%.
OEE = 0.867 x 0.846 x 0.970 = 71.1%.

Notice what the breakdown tells you that the single number doesn't. Quality looks fine. The two big levers are the 60 minutes of stops and the 60 minutes of speed loss hiding inside run time — the machine ran, but at an average pace well below ideal. Those speed losses are usually micro-stops and slow cycles nobody logs, which is why Performance is the factor most shops are most surprised by.

A common cross-check: OEE also equals (Ideal Cycle Time x Good Count) / Planned Production Time. Here: (0.5 x 640) / 450 = 320 / 450 = 71.1%. If your three-factor math and this 'fully productive time' math disagree, one of your inputs is wrong.

What counts as a good OEE number

The figure of 85% is often cited as world-class OEE for discrete manufacturing, and many real-world shops measuring honestly for the first time land somewhere around the 40-60% range. Treat both as folklore benchmarks rather than targets handed down from anywhere authoritative — the honest answer is that a good OEE number is one that's measured consistently and trending up.

Comparing OEE across different machines, products, or plants is where the metric gets abused. A slow, stable machine running one product all week will post a better OEE than a flexible machine absorbing ten changeovers, and that says nothing about which is managed better. Use OEE to compare a machine against its own history, and use the three-factor breakdown — not the headline number — to decide what to fix.

Mistakes that make OEE lie

Padded ideal cycle times — Performance over 100% is the giveaway. Fix the ideal rate, don't celebrate.
Quietly reclassifying unplanned downtime as 'planned' to protect Availability. Decide the planned/unplanned rules once, write them down, and don't move them shift to shift.
Counting reworked parts as good. Rework consumed capacity; strict OEE treats it as a quality loss.
Ignoring micro-stops because each one is under a minute. Thirty 40-second jams per shift is 20 minutes of loss that appears nowhere unless Performance is calculated from actual counts.
Only tracking OEE on the plant's best machine. The constraint machine's OEE is the one that gates output — measure there first.
Chasing the aggregate number instead of the factor. An OEE that drops from 71% to 68% is a prompt to look at which factor moved, not a verdict.

Tip Log every shift, even the ugly ones — especially the ugly ones. A month of honest daily numbers gives you a trend line and a loss breakdown; a few cherry-picked good shifts give you decoration.

Common questions

Do scheduled breaks count against OEE?

Not if you define planned production time to exclude them, which is the common convention. What matters is consistency: pick a definition of planned time, document it, and apply it every shift. (A related metric, TEEP, measures against all calendar time instead — useful for capacity planning, but it's a different question.)

Should changeovers count as downtime?

In most OEE conventions, yes — changeovers are availability loss, because the machine isn't producing. That feels harsh, but it's deliberate: it keeps changeover reduction visible as an improvement lever. If you exclude changeovers, say so explicitly wherever you report the number.

How is Performance ever over 100%?

It means the machine produced faster than your stated ideal cycle time — which means the ideal cycle time is set too slow. Update it to the true best rate; otherwise Performance masks losses and your OEE is inflated.

Do I need sensors and an MES to track OEE?

No. Per-shift OEE needs six numbers: shift length, planned exclusions, unplanned stop minutes, total count, good count, and ideal cycle time — all of which an operator can record on paper or a tablet in a minute at shift end. Automated data collection adds resolution, not validity.