To calculate OEE (overall equipment effectiveness), multiply three ratios for the same period: availability (running time ÷ planned production time), performance (actual output ÷ what the ideal rate would have produced in the running time) and quality (good units ÷ total units started). The result equals good units divided by what the line could have made at its ideal rate across all planned time. A line that is available 80% of the time, runs at 85% of ideal speed and makes 97% good parts has an OEE of about 66%.
The arithmetic is simple. The definitions are not. The Production Equation takes one invented line through one week and shows that, without anything changing on the factory floor, it can honestly report an OEE of 64.6%, 78.6% or 85.7%, depending on how three terms are defined. That gap is why OEE is valuable for improving one line over time and nearly useless for comparing two companies whose definitions are unknown.
What is the OEE formula?
OEE = Availability × Performance × Quality
- Availability = running time ÷ planned production time. It captures breakdowns, changeovers and other stoppages.
- Performance = units started ÷ (running time × ideal rate). It captures slow running and minor stops.
- Quality = good units ÷ units started. It captures scrap and, depending on the definition, rework.
Multiply them and the running time and units cancel out, so OEE also equals good units ÷ (planned time × ideal rate). That shortcut is the best check on your arithmetic.
How to calculate OEE step by step
Take the worked example from Chapter 4 of The Production Equation. The figures are invented.
A line is scheduled for 80 hours in a week: two eight-hour shifts for five days. It loses 4 hours to planned maintenance, 6 to changeovers and 8 to unplanned stoppages, so it runs for 62 hours. Its design rate is 120 units an hour. It starts 6,800 units. Of those, 6,200 pass every test the first time, 400 are reworked and then pass, and 200 are scrapped.
Step 1: Define planned production time
Decide what counts as time the line was supposed to run. The strict choice is all scheduled time: 80 hours. Write the choice down. Every later comparison depends on it.
Step 2: Calculate availability
Running time divided by planned time: 62 ÷ 80 = 77.5%.
Step 3: Calculate performance
Units started divided by what the design rate would have produced in the running time: 6,800 ÷ (62 × 120) = 6,800 ÷ 7,440 = 91.4%.
Step 4: Calculate quality
Good units divided by units started. On the strict definition only first-pass good units count: 6,200 ÷ 6,800 = 91.2%.
Step 5: Multiply
0.775 × 0.914 × 0.912 = 64.6%. Check it with the shortcut: 6,200 good units ÷ (80 hours × 120 an hour = 9,600) = 64.6%.
Why does the same line report different OEE numbers?
Now change three definitions, each of which is in use somewhere, without changing anything that happened on the line.
| Definition used | Availability | Performance | Quality | OEE |
|---|---|---|---|---|
| All scheduled time; design rate; first-pass good units only | 77.5% | 91.4% | 91.2% | 64.6% |
| Maintenance and changeovers excluded from planned time; reworked units counted as good | 88.6% | 91.4% | 97.1% | 78.6% |
| As above, with a "standard" rate of 110 an hour instead of the design rate | 88.6% | 99.7% | 97.1% | 85.7% |
Stylized example from Chapter 4 of The Production Equation; invented numbers.
Excluding the 10 hours of maintenance and changeovers shrinks planned time to 70 hours and lifts availability to 88.6%. Counting the 400 reworked units as good lifts quality to 97.1%. Measuring performance against a "standard" rate of 110 an hour rather than the 120 the equipment was designed for lifts performance to 99.7%. The third row now beats the 85% often quoted as world-class, for a line whose strict OEE is below 65%.
None of these definitions is a lie. The lesson is that OEE is a measure for managing a process over time, defined by the people who report it. Before you compare lines, plants or suppliers, get their definitions first. The book calls this the evidence problem: company-defined measures such as OEE, yield and capacity sit below audited and regulator figures on the ladder of trust.
What is OEE worth in dollars?
The reason to improve OEE is that it is usually the cheapest capacity a manufacturer can buy. The book's Appendix A follows an invented line with 4,000 planned hours a year, an ideal rate of 60 units an hour, a $120 price and $80 of variable cost per unit, so each good unit contributes $40. The market will take at most 200,000 good units a year.
| Case | Availability | Performance | Quality | OEE | Good units a year | Contribution a year | Change vs baseline |
|---|---|---|---|---|---|---|---|
| Baseline | 80% | 85% | 97.0% | 66.0% | 158,304 | $5.94m | — |
| Planned maintenance | 88% | 85% | 97.0% | 72.6% | 174,134 | $6.53m | +$0.59m |
| Quicker changeovers, fewer minor stops | 88% | 92% | 97.0% | 78.5% | 188,475 | $7.07m | +$1.13m |
| Better first-pass quality | 88% | 92% | 99.5% | 80.6% | 193,332 | $7.66m | +$1.72m |
| "World-class" benchmark | 90% | 95% | 99.9% | 85.4% | 200,000 sold | $7.98m | +$2.04m |
Stylized example from Appendix A of The Production Equation; invented numbers. Scrapped units lose their variable cost.
Planned maintenance, quicker changeovers and better first-pass quality lift OEE from 66.0% to 80.6% and contribution by $1.72 million a year, with the same machines, people and fixed costs. The last step adds only $0.33 million more, because output now exceeds what the market will take. Beyond demand, better OEE saves cost, such as a shift, overtime or a second line not bought, rather than earning revenue. Because fixed costs do not change, every dollar of extra contribution is a dollar of operating profit, which is the operating leverage of a factory at work.
How to calculate the cost of downtime
The same model prices an hour of unplanned downtime, and the answer depends on one question: is the line sold out?
On a sold-out line, a lost hour is lost sales. At baseline performance and quality, the line would have made about 49.5 good units in that hour. The hour costs their contribution plus a crew of six standing idle at a loaded $35 an hour: about $2,189.
On a line with slack, the output is made up later on overtime. The hour costs only the crew's wages at 1.5 times the normal rate: about $315.
To do this on your own line, take good units per hour at current performance and quality, multiply by contribution per unit, and add idle labor if the line is the bottleneck. If it is not the bottleneck, price the overtime or extra shift needed to recover the output instead. A plant's output is set by its constraint, so raising OEE on a line that is not the constraint saves cost but sells nothing more.
Common OEE mistakes to avoid
Changing the definition mid-program. Excluding changeovers from planned time halfway through a year produces an improvement nobody earned. Keep one definition and trend it.
Using nameplate or "standard" rates loosely. Measure performance against a rate the line was designed to sustain, and record which one you used. A standard set from past experience hides slow running.
Counting rework as good. Rework consumes labor and capacity. If you count it as good for OEE, track first-pass yield separately. Over several steps the difference compounds: a process with four steps at 98%, 97%, 99% and 96.5% first-pass yield has an average step yield of 97.6% but a rolled throughput yield of only 90.8%.
Comparing across companies. Two plants' OEE figures can be compared only after their definitions are. Vendor claims that software or AI "improved OEE by 20%" deserve the same questions: what baseline, whose definition, how many lines, and what else changed.
Ignoring the cost of the improvement. Training, spare parts, quick-change tooling and inspection all cost money. The dollar value of an OEE gain is net of them.
Where OEE fits in a production system
OEE is an output of process capability, not a substitute for it. The Toyota Production System attacks the three terms directly: jidoka and stopping at the source of defects raise quality, quick changeovers raise availability, and standard work and kaizen raise performance. Automation can help too, but the book's chapter on automation and the factory floor shows that a line of machines in series is only as available as the product of their uptimes. Ten machines that each run 98% of the time give a line that runs about 82% of the time.
Tesla learned that in 2018, when its highly automated Model 3 lines in Fremont could not reach rate. It removed robots, added people and built a third assembly line under a tent before reaching 5,000 cars a week in the last week of June 2018, six months late. Industrial-automation suppliers such as Rockwell Automation and Siemens sell the controls and software that collect OEE data, and the book's section on automation, robots and the smart factory covers what that data can and cannot prove.
OEE is one of the levers behind the broader economics described in the guide to how manufacturing companies make money: better conversion and fuller fixed assets.
The takeaway
Calculating OEE takes three ratios and one multiplication; using it well takes a stable definition and the discipline to compare a line with itself rather than with a competitor's brochure. On the book's invented line, maintenance, changeovers and first-pass quality added $1.72 million a year of contribution with no new equipment, which is why OEE is usually the cheapest capacity a manufacturer can buy. Its value depends on whether the line is sold out, so price downtime as lost sales only on the true bottleneck. The worked examples on measurement come from Chapter 4 of The Production Equation and its Appendix A, which you can preview on the book page.
Frequently asked questions
What is a good OEE score?+
An OEE of about 85% is often quoted as world-class, typically as 90% availability, 95% performance and 99.9% quality. The figure is only meaningful once you know how planned time, ideal rate and good units were defined.
Can OEE be higher than 100%?+
Not if it is defined correctly against the equipment's true ideal rate. A reading above 100% usually means the performance term uses a standard rate lower than the line can actually run.
Should planned maintenance count against OEE?+
Conventions differ. Counting all scheduled time gives a stricter figure; excluding planned maintenance and changeovers raises availability without changing output. Pick one definition and keep it stable over time.
What is the difference between OEE and first-pass yield?+
First-pass yield is the share of units that pass every test the first time without rework. It corresponds to the quality term inside OEE, which also multiplies in availability and performance.
Go deeper in The Production Equation
Chapters
Industry studies
From the research behind The Production Equation
How Manufacturers Create, Capture and Destroy Value
25 chapters, 24 industry studies, 100 companies and 844 strategic moves.
See the bookEducational content, not investment, legal, tax, medical or engineering advice. Figures marked “reported” come from filings and published sources that have not been independently audited.