OEE (overall equipment effectiveness) is calculated by multiplying three ratios: OEE = Availability × Performance × Quality. Availability shows how much of the planned time the machine actually ran, performance shows how close it came to its ideal speed while running, and quality shows how many of the parts it made were right first time. You can reach the same result by a shortcut: good count × ideal cycle time ÷ planned production time.
What OEE measures, and why it matters
It is easy to say a machine "ran all day". OEE shows, in a single percentage, how true that statement is. An OEE of 100% means the machine ran for the whole of its planned time, at ideal speed, without producing a single defective part. No line ever gets there; what matters is seeing where the loss is.
OEE's strength is that it breaks down into three components. Two machines can have the same OEE, yet one suffers from frequent breakdowns while the other runs slowly or produces scrap. That breakdown tells you whether maintenance, process engineering or quality needs to act. If the loss comes from frequent failures, the natural next step is moving to predictive maintenance; if quality is low, it is process capability with SPC and Cpk.
What losses cost: why measure at all?
The biggest single drag on OEE is unplanned downtime, and it is expensive:
Siemens estimates that the world's 500 largest companies lose almost $1.4 trillion a year to unplanned downtime, equivalent to 11% of their revenues. (Siemens, The True Cost of Downtime 2024)
According to the same report, an hour of unplanned downtime costs $2.3 million in a large automotive plant and $36,000 in fast-moving consumer goods (FMCG) plants, and large plants average 25 unplanned downtime incidents and 27 lost production hours a month. Those figures relate to large companies, but the logic holds at any scale: you cannot reduce downtime you do not measure.
There is a widely quoted benchmark for setting targets. According to OEE.com's World-Class OEE page, a "world-class" OEE is often taken to be 85%, derived from goals of 90% availability, 95% performance and 99% quality in discrete manufacturing. The same source notes that most manufacturers in reality run closer to 60%. The benchmark goes back to Seiichi Nakajima's 1984 book on TPM and applies to discrete rather than process manufacturing, so the most useful target for your plant is improvement on your own baseline.
The OEE formula: three components
| Component | Formula | Loss it reveals |
|---|---|---|
| Availability | Run time ÷ Planned production time | Breakdowns, setups and changeovers, waiting for material |
| Performance | (Ideal cycle time × Total count) ÷ Run time | Minor stops, reduced speed |
| Quality | Good count ÷ Total count | Scrap, rework, start-up losses |
| OEE | Availability × Performance × Quality | All losses combined |
Three definitions deserve care:
- Planned production time is shift length minus planned breaks and any time when no production is scheduled. Decide once whether planned maintenance is excluded here, and apply that rule consistently.
- Ideal cycle time is the fastest time the machine can theoretically achieve for that product, not "the speed we usually run at".
- Good count means parts that meet the quality specification first time. Parts that are reworked and saved do not count as good.
A worked example, step by step
Take a single shift on a CNC machine:
- Shift length: 8 hours = 480 minutes. Planned breaks: 30 minutes. Planned production time = 450 minutes.
- Stoppages: 35 minutes of breakdown and 12 minutes of tool change, 47 minutes in total. Run time = 450 − 47 = 403 minutes.
- Availability = 403 ÷ 450 = 89.6%
- Output: ideal cycle time is 30 seconds (0.5 minutes), and 700 parts were produced in total. Ideal time = 700 × 0.5 = 350 minutes.
- Performance = 350 ÷ 403 = 86.8%
- Quality: 28 of the 700 parts were defective and 672 were good. Quality = 672 ÷ 700 = 96.0%
- OEE = 0.896 × 0.868 × 0.960 ≈ 74.7%
Check it with the shortcut: 672 good parts × 0.5 minutes = 336 minutes of fully productive time; 336 ÷ 450 = 74.7%. When both routes give the same answer, your data is consistent.
In this example the biggest loss is performance: of the 403 minutes it ran, the machine lost about 53 minutes to minor stops and slow running. On a paper shift form that loss usually does not appear at all, because nobody writes down two-minute jams. An andon system that flags each stop the moment it happens both records these jams and shortens the response time.
The six big losses
The three OEE components map onto what TPM (total productive maintenance) calls the "six big losses":
| Component | Loss | Example |
|---|---|---|
| Availability | Breakdowns | Motor, hydraulic or electrical failure |
| Availability | Setups and adjustments | Die change, waiting for first-off approval |
| Performance | Minor stops and idling | Jams, sensor faults, gaps in material feed |
| Performance | Reduced speed | Running slower because of a worn tool |
| Quality | Process defects | Out-of-tolerance parts, rework |
| Quality | Start-up losses | Scrap made while warming up or adjusting |
Recording downtime reasons against these categories turns OEE from a report card into an improvement tool.
How to collect OEE data
The formula is simple; the hard part is making sure the numbers going into it are right. If you are starting by hand, a plain shift form that captures the following is enough:
- Planned production time, and the planned breaks deducted from it
- Start and end time of every stoppage, with its reason (breakdown, setup, waiting for material, no operator and so on)
- Product code and that product's ideal cycle time
- Total parts produced and defective parts, with the type of defect
Use a short, predefined list of downtime reasons rather than free text. "Machine stopped" is useless for analysis; concrete categories such as "die change", "waiting for raw material" or "sensor fault" can still be compared months later.
The limit of manual measurement is minor stops. Expecting operators to log every two-minute jam is unrealistic, so manually calculated performance almost always comes out too high. A run/stop signal and an automatic part counter taken from the machine close that gap (even machines without a PLC can be handled by collecting data from legacy machines): the system measures how long the stop lasted and the operator only selects the reason. This is where plants that start measuring OEE often get their first surprise, discovering that what they had long called "breakdowns" is really a stream of short, frequent stoppages.
Five common mistakes
- Setting a loose ideal cycle time. If "normal speed" is treated as ideal, performance looks artificially high and the real loss stays hidden.
- Not recording minor stops. Stops of a few minutes vanish from manual records, yet added up they are often the largest loss.
- Counting reworked parts as good. Quality looks better than it is, and the cost of rework becomes invisible.
- Comparing different machines on raw OEE. Where product mix and process differ, a bare OEE comparison misleads; look at the components.
- Turning OEE into an operator scorecard. Once the number is used to punish, data quality suffers. OEE exists to find the loss.
How Digital Bridge approaches it
You can calculate OEE by hand, but to be reliable and sustainable the data has to come from the machine automatically. We build that up step by step:
- We take data from the machine. Run/stop signals and part counts come from the machine or its PLC; on older equipment, industrial data terminals let operators select the downtime reason.
- We calculate OEE live inside the MES. In our MES production management solution, availability, performance and quality are tracked live for each machine; losses are classified into categories such as breakdown, setup, speed loss and quality reject, and the end-of-shift report is generated automatically. The MES integrates with Logo, SAP, Mikro and custom ERP systems.
- We make scrap traceable. With production traceability and barcoding, you can see which batch and which machine a defective part came from.
- We tackle breakdown losses at the root. Where availability loss comes from failures, we run a predictive maintenance pilot on one to three critical machines and show the value with metrics within the first 30–60 days.
- We give managers the right screen. A BI dashboard brings the OEE trend, the machine with the most downtime and the largest loss category together on one screen.
Our Industry 4.0 roadmap for SMEs explains where to begin; measuring OEE is the second stage on that roadmap.
How measured OEE feeds work orders and the ERP is covered in MES vs ERP, and scheduling against real capacity in production scheduling with APS; availability and performance rates are direct inputs to planning.
Next step
Pick the machine that limits your capacity and apply the seven steps from the example above by hand for one week. When minor stops and the ideal cycle time are entered honestly, the result is usually lower than expected, and that gap is your improvement opportunity. Then contact us and, during a free site assessment, we will look together at how to set up automatic OEE measurement on that machine.