Where does your plant stand? This page keeps only the OEE figures that have a public source you can open and check, with the link next to each one. Where no published source exists, for example a typical OEE for one industry, we say so instead of printing a guess.
The headline numbers
What each industry loses most to
Published figures by industry are scarce: one study of 23 Swedish companies found the food and beverage group averaging 74% and other automated discrete production 59%, measured automatically in a system where most machines reported the default 100% for speed and quality (source: Hedman et al., 2016). So this table names the losses, not a number. For every figure we could trace, see the OEE values reported in published studies. Compare your line with its own trend first, then with the world-class and typical figures above.
| Industry | Dominant loss |
|---|---|
| Food & Beverage | Changeovers, CIP, filler/capper micro-stops |
| Automotive | Line-stop cascades, extreme cost per stop |
| Packaging & Converting | Micro-stops, web breaks, line imbalance |
| Plastics / Injection Moulding | Cycle-time drift, mould changeovers |
| Pharmaceuticals | Validation, cleaning, documentation, serialization |
| Electronics & SMT | Feeder/nozzle micro-stops, placement & solder defects |
| CPG / Consumer Goods | SKU & promo changeovers, packaging micro-stops |
| Metals & Foundry | Furnace/melt downtime, die changes, casting yield |
| Machinery & Fabrication | Setup-heavy high-mix, tooling, waiting |
| Aerospace & Defense | Setups, hard-alloy tool wear, inspection & traceability |
| Medical Devices | Validation, documentation, micro-stops, quality holds |
| Textiles & Nonwovens | Yarn/web breaks, changeovers, speed losses |
| Paper & Pulp | Web breaks, grade changes, clothing/roll changes |
| Building Materials & Cement | Kiln/crusher/mill availability, abrasive wear |
Read the 85% with care. It is a target, not an average: published plant studies found average OEE of about 55% and 65% (sources: Ljungberg, 1998; Hedman et al., 2016). The higher figure comes from automatic data in which most machines reported the default 100% for speed and quality, so it flatters. For a high-mix or changeover-heavy plant, your own trend is the better yardstick.
Downtime cost figures, each with its source, are on the downtime statistics page.
World-class depends on the type of process
The single 85% figure fits batch and discrete lines. R. C. Hansen, building on Nakajima's TPM figures, sets the world-class bar by type of process and grades key processes in three bands (source: Hansen, Overall Equipment Effectiveness, Industrial Press, 2001):
| Type of process | World-class OEE |
|---|---|
| Batch and discrete processes | 85% or more |
| Dedicated discrete, and non-dedicated continuous processes | 90% or more |
| Dedicated continuous processes | 95% or more |
| OEE on a key process | Hansen's reading |
|---|---|
| Below 65% | Unacceptable: money is leaking now, act on it |
| 65% to 75% | Passable, but only if the quarterly trend is improving |
| 75% to 85% | Pretty good: keep moving toward the target for your type of process |
Two cautions from the same book. Compare OEE only where it is calculated the same way: the same ideal rate, the same rule for what counts as a good unit, and the same rule for which time is planned. And apply the benchmark to the bottleneck first; a high OEE on a machine that waits on its neighbour adds no output. Check your own figure the way the book recommends, from good units, with the OEE calculator.
Reader benchmark
Reader-reported OEE will show here, with the number of plants behind each figure, once storage for reader numbers is switched on and enough plants have sent theirs. Nothing here is estimated.
Loading reader numbers...
Calculate your own OEE in two minutes, then come back and send it to the team.
Use this benchmark on your site
Writing about OEE? Drop this benchmark into your article. It is free to use, and it credits OEE Lab with a link so your readers can find the full data.
Renders as a compact, light-theme card with the headline numbers, their source and a link to this page.
Add your plant's numbers
Send your industry and OEE to the OEE Lab team with your work email. It takes 30 seconds.
FAQ
What is a good OEE score?
The long-cited world-class figure is 85%, traced to Nakajima, Introduction to TPM (1988). Published studies of real plants report averages of about 55% (Ljungberg, 1998) and 65% (Hedman et al., 2016), so most lines have a wide gap to close.
Why is there no typical OEE per industry on this page?
Because we could not find a public source that measures it. We print a figure only when we can open its source and find the number in it.
How do I compare my own line with this benchmark?
Run one shift through the OEE calculator, which compares your result with the sourced figures, then add your plant's numbers in the survey above.
Methodology
Every figure on this page links to the page it comes from, with the date we read it. We print a figure only when we could open the source and find the number in it; when we could not, for example for a typical OEE by industry, the figure is left out. Numbers sent through the form above are not published on this page until storage for reader numbers is switched on.
Cite as: OEE Lab. (2026, October 3). The 2026 manufacturing OEE and downtime benchmark. https://oeelab.com/oee-benchmark-report. Please credit the original study too; each figure links to it.
Closing the gap to world-class
Between a typical 60% (the mid-point of the published averages of about 55% in Ljungberg, 1998 and 65% in Hedman et al., 2016) and world-class 85% (traced to Nakajima, Introduction to TPM, 1988) sit 25 points of output, and short stops that nobody logs are among the hardest to close by hand. For closing it, we recommend Fabrico: video of a stop streamed on request next to its machine data (where cameras are fitted), OEE read live from PLCs or IoT sensors, every stop as minutes of OEE lost per line and shift, EU-built, with EU customers hosted on AWS in Frankfurt, and ISO 27001 / 20000-1 / 9001 (supports audit-readiness).
Book a Fabrico demo