If you want to improve OEE, you need machine data rather than estimates from the shift log. Overall Equipment Effectiveness is the product of three factors: availability, performance, and quality. Each factor can only be accurately calculated if a machine actually reports when it is idle, how fast it is running, and how much scrap it produces.
OEE (Overall Equipment Effectiveness) is calculated as the product of availability, performance, and quality, each expressed as a percentage. The world-class benchmark of 85 percent applies to highly automated mass production, not to every industry. Who Switched from manual to automatic data entry, OEE usually drops at first because previously unnoticed micro-downtimes become apparent for the first time.
Improving OEE starts with machine data
Availability compares the actual runtime to the planned production time. Performance compares the actual cycle time to the ideal cycle time. Quality compares the number of good parts to total production. Multiplying all three results in a number between zero and one hundred percent that can be immediately interpreted even without specialized knowledge of setup times.
That is the value of the metric. A production manager compares two lines without knowing the details of either one. Aggregation becomes a problem as soon as one of the three factors is based on poor data—such as an estimated setup time or a microbreak that was never logged. This article describes how data can be tracked with consistent quality from the sensor all the way to the dashboard. From Sensor to Dashboard: The Data Chain in Plant Monitoring.
World-class, average, and low-volume production: realistic figures
The 85-percent benchmark originates from TPM literature from the 1980s and describes highly automated mass production with few product changeovers, such as in the chemical industry or in continuously operating plants. Those seeking to improve OEE should use this figure as a guideline, not as a mandatory target for every piece of equipment.
According to the provider TeepTrak In 2026, the median for German companies across all industries will be around 60 percent, and the upper quartile will be around 75 percent. The software provider Symestic cites a global average of 55 to 60 percent for 2025. Both figures come from manufacturers of tracking software; there are no independent official statistics on this. In discrete manufacturing, setup times and product variety structurally push the figures below the level seen in the process industry.
That is why comparing performance within one’s own industry matters more than comparing it to a general benchmark. A dairy plant with mandated cleaning cycles will never match the performance of a Konti system in the paper industry, no matter how well the shift works.
The world-class standard in TPM literature requires 90 percent availability, 95 percent performance, and 99 percent quality all at the same time.
Nakajima, *Introduction to TPM*, 1988
The "Six Big Losses" as a Starting Point
The losses contributing to OEE can be classified into six categories:
- Outages and disruptions that reduce availability
- Setup and adjustment times during product changeovers
- Very brief stoppages and idle time of less than about five minutes
- Reduced speed compared to the ideal cycle time
- Start-up losses after maintenance or a shift change
- Scrap and rework due to quality defects
The first two categories relate to availability, the middle two to performance, and the last two to quality. Anyone who wants to improve OEE should focus on the factor with the greatest leverage—in practice, this usually means addressing minor downtime, since it is the least frequently documented.
Why Automatically Tracked OEE Drops First
A shift log rarely records tasks that take less than five minutes. An operator who readjusts a material feed three times an hour does not record this on any form. Over the course of an eight-hour shift, these minor stoppages add up to a significant amount of time.
That is why the measured OEE almost always decreases as soon as a facility switches from manual record-keeping to automatic machine data collection. This is not a step backward. It is the first honest figure a plant has ever seen. The article explains what rights operators actually have regarding this machine data vis-à-vis the equipment manufacturer. EU Data Act: Access to Machine Data as a Requirement and an Opportunity.
The extent of the decline depends on how many minor stoppages have gone unrecorded so far. A figure that stood at 78 percent on paper often drops significantly below that level after the changeover, even though nothing has changed on the machine itself. Only now can OEE be improved, because the causes of the losses are now visible.
For machine data to be more than just a timeline, every downtime event must have a cause. Signals from the control system indicate that a machine is down, but not why. Only by linking this information with job data, tool changes, and PLC messages can the downtime be assigned to one of the “Six Big Losses” categories. A two-step approach has proven effective: The machine provides a timestamp and error code, and an operator confirms or corrects the suggested cause on a tablet. If the analysis is performed directly at the machine, the categories are available during the current shift. This article demonstrates how the necessary connection to the control system can be verifiably secured—even in compliance with NIS2. NIS2 in Production: Demonstrating OT Security.
Distinction: OEE is not TEEP
OEE always refers to scheduled production time. If a machine is idle on a Sunday because no shift is scheduled, this is not included in the OEE calculation. TEEP (Total Effective Equipment Performance), on the other hand, is calculated based on the full calendar time—that is, 24 hours a day, 365 days a year.
This distinction becomes important as soon as a company begins to consider adding additional shifts. An OEE of 80 percent in single-shift operation says nothing about how much capacity lies idle during the remaining 16 hours. TEEP makes this reserve visible and is therefore suitable for investment decisions, while OEE is suitable for daily shift management.
Why Treating OEE as a Purely Departmental Goal Creates Misaligned Incentives
If OEE is set as a target exclusively for production, predictable workarounds will occur. A team may set the ideal cycle time low in order to preserve the performance factor. Another team might schedule setup operations during unplanned periods so that they do not count as a loss of availability. Both approaches improve the metric, not production.
A more sensible approach is one in which production, maintenance, and quality assurance share responsibility, using the same raw data as a foundation. If maintenance is measured by unplanned downtime and production by an overall performance metric, both departments will work against each other to optimize their performance. Based on our own project experience, this is precisely where the greatest resistance to automated data collection lies. It brings to light conflicts that were previously masked by inaccurate figures. This article shows how the same raw data can also be used to reduce energy consumption. Energy Monitoring: 5 to 25 Percent Savings Through Transparency.
| Production Method | Typical range | Upper quartile |
|---|---|---|
| Discrete Manufacturing (Mechanical Engineering, Electronics) | 50 to 65 percent | 70 to 80 percent |
| Highly Automated Mass Production | 55 to 70 percent | 75 to 85 percent |
| Process industry, continuous | 75 to 90 percent | over 90 percent |
| Food and Beverages | 45 to 60 percent | 65 to 75 percent |
Industry benchmarks are reference values derived from vendor studies and project experience; they are not official statistics. For systems with very small batch sizes or one-off production, the classic OEE formula is of limited value because there is no stable ideal cycle time. In such cases, availability and quality, when considered individually, often provide more insight than the combined metric.
Identify the factors behind the metric
The first automatically measured OEE is almost always lower than the figure recorded in the shift log. Whether minor downtimes on your own equipment can be recorded and attributed to a specific cause depends on the available control data.
Frequently Asked Questions
What is a good OEE value to start with?
A good starting point is the first automatically measured figure itself, regardless of its value. Facilities measuring honestly for the first time often end up between 40 and 60 percent, because minor downtime and speed losses become visible for the first time. More important than the absolute value is the trend over the following months, measured using a consistent time frame. A plant that increases from 45 to 55 percent has achieved more than one with a high but inaccurate figure.
How often should OEE be analyzed?
For shift management, a daily update on the key performance indicator is sufficient to allow shift supervisors and maintenance staff to respond immediately to urgent problems. For comparisons between production lines or locations, time periods of three months or longer are more meaningful, because daily values fluctuate significantly due to individual disruptions and can otherwise lead to incorrect conclusions. Anyone who uses a single bad week as the basis for an investment decision is mistaking noise for a genuine trend.
Is MES software alone enough to improve OEE?
No. Software provides the data but does not change processes on its own. Without a set routine in which those responsible regularly review the numbers to determine specific actions, even the most accurate data collection remains merely a dashboard with no impact on the shop floor. Only when shift supervisors, maintenance, and quality assurance derive shared priorities from the same data does the way production actually takes place change. This holds true regardless of how advanced the software being used is.
torck develops custom software for industry and retail and connects machines, MES systems, and control stations into a seamless data flow all the way to the dashboard. Because the teams in Maxhütte-Haidhof, Vienna, and Rabat work with a time difference of no more than two hours, and a single contract with the German company torck GmbH covers the service, downtime tracking is implemented in a way that fits the existing plant. Learn more about the implementation at Plant Monitoring, regarding sensor connectivity in the article „LoRaWAN, 5G, or Wired: A Comparison.“.
This article refers to laws and regulations to put technical decisions in context. It is not legal advice. Whether and how a rule applies to your company is a question for your legal department or a law firm.
German 
English 
French