Performance
What is a good OEE score? Benchmarks and the six big losses
The OEE figure most often quoted as the benchmark is 85%, from Seiichi Nakajima's TPM work in Japan, but many plants measure much lower when they first count honestly. A good OEE score is one that rises steadily on the same machine, under the same counting rules. To move it, find which of the six big losses takes the most minutes and work on that first.
Where the 85% OEE benchmark comes from
OEE (overall equipment effectiveness) is the share of planned production time that made good parts at full speed. It multiplies three factors: Availability, Performance and Quality. If you need the formula or want to work out your own number, use the free OEE calculator first and come back.
The 85% figure goes back to Seiichi Nakajima, who led the work on TPM, OEE and the six big losses at the Japan Institute of Plant Maintenance. In Introduction to TPM (1984 in Japanese, 1988 in English) he gave minimum goals for each factor. They are usually quoted as:
- Availability: 90%
- Performance: 95%
- Quality: 99.9%
- OEE: about 85% (0.90 × 0.95 × 0.999)
Notice how uneven the factor goals are. Quality is almost perfect, Availability has the most room. That is because a reject is expensive in a way that a short stop is not.
Also notice where the numbers come from. Vorne Industries, which publishes OEE.com, points out that they were set in Japan, in the 1970s, mostly for automotive plants, and apply to discrete manufacturing rather than process plants. Vorne also writes that most manufacturers it works with sit closer to 60%. So 85% is a good long-term direction. It is a poor first target for a plant that has just started measuring.
OEE benchmarks: a rough guide you can copy
Use these bands to decide where to look first, not to grade your team. They are rules of thumb drawn from the sources below, not an industry standard, and the same score means different things on a single-product press and a high-mix CNC cell.
| Your OEE | What it usually says | Check first |
|---|---|---|
| Below 40% | Common when a plant first measures honestly, or when the rules are too strict | Are planned breaks and planned PM taken out of planned time? Is the ideal cycle time realistic? |
| 40% to 60% | Large, visible losses. Plenty to gain from basics | The two biggest losses by minutes. Usually breakdowns or changeovers |
| 60% to 75% | Basics are in place. Hidden losses start to matter | Minor stops and slow cycles that nobody logs |
| 75% to 85% | A well-run line by most yardsticks | Startup rejects, speed losses on specific products, setup time |
| Above 85% | Either very good, or the rules are too generous | Audit the inputs before you celebrate: ideal cycle time, what counts as planned time |
Whatever band you are in, compare a machine with itself over time. A jump of ten points after a change of counting rules is not improvement.
The six big losses, and how to log each one
Nakajima grouped equipment losses into six and tied each pair to one OEE factor. Knowing the score without knowing the losses is like knowing your temperature without knowing why. The TPM guide covers the background. This table is about capture: how each loss gets into your records on a normal shift, and who usually acts on it.
| Loss | OEE factor | How to capture it | Usual owner |
|---|---|---|---|
| Equipment failure | Availability | Breakdown logged against the machine with start time, end time and cause | Maintenance |
| Setup and adjustment | Availability | Changeover start to first good part, per product change | Production and setup crew |
| Idling and minor stops | Performance | Short stops tallied by the operator, or counted from sensor signals | Production with maintenance |
| Reduced speed | Performance | Actual count against ideal cycle time. Any gap not explained by stops | Process engineering |
| Startup rejects | Quality | Rejects from warm-up and the first parts after a start or changeover | Production and quality |
| Process defects | Quality | Scrap and rework during steady running, by defect type | Quality |
Minor stops and reduced speed are the two that most plants never log. They hide inside Performance, and they are often bigger than the breakdowns everyone talks about.
Worked example: a CNC turning cell in Chakan
Worked example (illustrative)
Take a CNC turning cell at an auto component supplier near Chakan, Pune. It runs two shifts of 480 minutes, six days a week, with a 30 minute meal break per shift. Planned production time for the week is 6 × 2 × 450 = 5,400 minutes. The cell makes ten variants of a shaft, so it changes over often. The plant head has put an 85% OEE target on the wall because it appeared on a slide. All numbers below are illustrative.
In the week the team logs 310 minutes of breakdowns and 16 changeovers of about 40 minutes each, 640 minutes in total. The ideal cycle time is 1.5 minutes per part. The cell makes 2,500 parts, and 2,410 are good. The operators keep a tally of short stops, which adds up to about 380 minutes.
| Item | Calculation | Result |
|---|---|---|
| Run time | 5,400 - 310 - 640 | 4,450 min |
| Availability | 4,450 ÷ 5,400 | 82.4% |
| Performance | (1.5 × 2,500) ÷ 4,450 = 3,750 ÷ 4,450 | 84.3% |
| Quality | 2,410 ÷ 2,500 | 96.4% |
| OEE | 0.824 × 0.843 × 0.964 | 66.9% |
| Setup and adjustment | 16 changeovers | 640 min |
| Minor stops | Operator tally | 380 min |
| Reduced speed | 4,450 - 3,750 - 380 | 320 min |
| Equipment failure | Logged breakdowns | 310 min |
| Startup rejects | 55 parts × 1.5 | 82.5 min |
| Process defects | 35 parts × 1.5 | 52.5 min |
What the example shows (illustrative). The six losses add up to 1,785 minutes, and 2,410 good parts × 1.5 minutes is 3,615 fully productive minutes. Together they make the 5,400 planned. OEE is 66.9%, which sounds bad against 85%. But the biggest loss is not breakdowns, it is changeovers. If setup work brings each changeover from 40 to 25 minutes, the cell gets back 240 minutes. At the same speed and quality that is about 240 × 0.843 × 0.964 ≈ 195 productive minutes, or 3.6 points of OEE. A believable three month target is about 70%, with changeover time as the measure the team watches every day.
How to set an OEE target your team will believe
- Fix the counting rules first. Write down what is planned time, what is a stop, what ideal cycle time you use and what counts as a good part. Use the same rules on every line.
- Measure for four weeks before you set anything. One good week proves nothing. You want a baseline you trust.
- Rank the six losses by minutes, not by how loud the complaints are. Minutes turn into OEE points directly. Percentages hide the size.
- Pick one or two losses and set a target on them. Changeover minutes, breakdown minutes or minor stops per shift. People can act on these.
- Set the OEE target as the result. Work out what the loss target is worth in points, as in the example, and aim for that within three or four months.
- Review weekly, reset quarterly. When the target loss is under control, move to the next longest bar.
A target works when the people on the line think they can hit it and can see what to change. An 85% sign on the wall does neither. Use this sequence instead:
Common mistakes
- Copying 85% onto every line. A line that makes one product all day and a cell that changes over sixteen times a week will never have the same OEE. Set targets per line, from that line's own baseline.
- Comparing plants with different rules. One plant takes planned maintenance out of planned time, another counts it as downtime. Their scores are not comparable until the rules match.
- Setting the ideal cycle time from today's average. If the ideal is too slow, Performance looks fine and reduced speed disappears. Use the design rate or the best proven rate.
- Averaging OEE percentages across machines. Add the minutes and counts first, then calculate the plant figure.
- Logging only breakdowns. Then Availability is the only loss anyone sees, and maintenance gets blamed for a score that minor stops and changeovers are pulling down.
- Treating OEE as a maintenance score. Maintenance owns a large share of Availability, but setups, speed and rejects belong to production, process and quality too. The OEE guide for maintenance leaders shows which part is maintenance's to fix.
Sources
- World class OEE (opens in a new tab), OEE.com (Vorne Industries)
- Six big losses (opens in a new tab), OEE.com (Vorne Industries)
- What is TPM? (opens in a new tab), Japan Institute of Plant Maintenance (JIPM)
- Overall equipment effectiveness (opens in a new tab), Wikipedia
Frequently asked questions
What is a good OEE score?
The benchmark most often quoted is 85%, from Seiichi Nakajima's TPM work. Many plants measure well below that when they first count honestly. A good score is one that rises steadily on the same machine under the same counting rules.
Is 85% OEE realistic for an Indian plant?
It can be, on a stable line with few changeovers and good basics. For most plants it is a long-term direction. Set a first target from your own four-week baseline and the biggest loss you plan to cut.
What are the six big losses in OEE?
Equipment failure and setup and adjustment reduce Availability. Idling and minor stops and reduced speed reduce Performance. Startup rejects and process defects reduce Quality.
Why is my OEE so much lower than my availability?
OEE multiplies Availability, Performance and Quality, so the losses stack. A line at 90% on all three factors has an OEE of about 73%.
Should every line have the same OEE target?
No. A high-mix line loses more time to changeovers than a single-product line. Set each target from that line's own baseline and compare a line with itself over time.
Keep reading
OEE software
OEE per machine from logged breakdowns and shift logs, with a worst-asset ranking.
ReadUsing OEE losses to choose maintenance work
Which part of the score maintenance can move, and how to pick the next job.
ReadOEE calculator
Work out Availability, Performance, Quality and loss minutes for one machine.
Read