The performance rate is the OEE factor that measures speed. It is calculated as ideal cycle time × total count ÷ run time, the ratio of how fast the machine actually ran to how fast it should have run. A correctly configured performance rate can never legitimately exceed 100%.
Of the three OEE factors, performance is the one plants most often get wrong, and almost always for the same reason: a soft ideal cycle time. Set the ideal too slow and performance inflates, sometimes past 100%, hiding real speed loss behind a flattering number. Set it too fast and the line looks broken when it is fine. This guide walks through the formula, a worked example, why performance can read over 100% and what that tells you, and how to set an ideal cycle time you can defend.
What is the performance rate?
The performance rate answers one question: when the machine was running, how close to its rated speed did it run? It is the middle factor of the OEE calculation sitting between availability (was the machine running at all?) and quality (were the units good?). Performance isolates speed, and only speed, it counts every unit the machine made, good or bad, because a defective unit still took the same cycle to produce.
That scope is the part people miss. Performance uses total count, not good count, precisely because it is measuring the machine's pace, not its output quality, quality losses are handled by the quality factor so they are not double-counted. A line running at full rate while making scrap has excellent performance and terrible quality, and OEE keeps those two stories separate on purpose. Performance is the pace story alone.
What the performance rate captures maps directly onto two of the six big losses: reduced speed and small stops. Both drag the same factor, because both mean the machine made fewer units than its run time allowed, one by running slow, the other by pausing briefly. The performance rate is the single number that measures their combined effect; splitting them back apart is a separate analysis. For the calculation itself, all that matters is that both show up as a gap between actual and ideal pace, and that gap is exactly what the formula quantifies.
How do you calculate the performance rate?
Multiply the ideal cycle time by the total count, then divide by the run time. Run time, also called operating time, is planned production time minus all downtime, so it is the time the machine was actually running.
The worked example: a machine with an ideal cycle time of 0.5 seconds per unit made 19,800 units (good and bad) across 11,000 seconds of run time. Performance = 0.5 × 19,800 ÷ 11,000 = 90%. Read the other way, the machine could ideally have made 22,000 units in that run time (11,000 ÷ 0.5), and it made 19,800-90% of the ideal. The two forms are algebraically identical; use whichever makes the number easier to explain on the floor.
How do you calculate performance correctly, step by step?
The formula is trivial; the inputs are where accuracy is won or lost. Work them in order:
- Nail down the ideal cycle time. The theoretical minimum time per unit at full rate. This is the single most important and most abused input, so treat it as a decision, not a guess.
- Count every unit, good and bad. Total count includes rejects, because performance measures pace, not quality. Excluding scrap here understates the machine's real speed and quietly steals loss from the quality factor.
- Measure run time, not scheduled time. Run time is planned production time minus all downtime. Using scheduled time instead double-counts availability loss inside performance.
- Keep the units consistent. If ideal cycle time is in seconds per unit, run time must be in seconds. Mismatched units are the most common arithmetic error in the whole calculation.
- Compute and sanity-check against 100%. Multiply ideal cycle time by total count, divide by run time. If the result exceeds 100%, stop, the ideal cycle time is wrong, not the machine miraculous.
- Cross-check the ideal against nameplate. Compare your ideal cycle time to the equipment maker's design rate. A large gap in either direction is a flag to re-derive it before trusting the performance number.
Why can the performance rate go over 100%, and what does it mean?
Performance over 100% means the machine produced more units than your ideal cycle time says was possible, which is impossible if the ideal is truly the fastest the machine can go. So the reading is not a triumph; it is a defect report on your ideal cycle time. The ideal was set too slow, and the "excess" performance is simply the machine running faster than a target that was too conservative to begin with.
The damage runs deeper than a nonsense number. A too-slow ideal does not just occasionally break 100%, it suppresses real speed loss on every run, because the line is always being graded against a target that is easy to beat. A plant can sit at 96% performance, feel good, and be leaving a fifth of its rate on the table, all because the ideal was set to a comfortable historical average instead of the true capability. This is exactly the speed loss that performance loss analysis is meant to surface, and it stays invisible until the ideal is honest.
How do you set an honest ideal cycle time?
Base it on demonstrated capability, not comfort. The defensible ideal cycle time is the fastest rate the process has actually sustained over a meaningful, stable run of good product, not a one-off record spike, and not a padded average of ordinary days. Cross-check that observed best against the equipment manufacturer's design or nameplate rate; the two should be in the same neighborhood, and a wide gap means one of them needs investigating before you trust it.
Hold the ideal fixed once set, and change it only with a documented reason, a genuine equipment upgrade, a product spec change, a corrected error. Quietly editing the ideal to make performance look better is the fastest way to make OEE meaningless, because it moves the very yardstick every trend is measured against. A stable, honest ideal is what lets you compare this week to last, this shift to that one, and this line to its twin. The same discipline underlies the cycle time measurement the ideal is drawn from.
What is a good performance rate, and how is it defined?
The common world-class reference for performance is 95% the middle term of the 90% availability × 95% performance × 99% quality split that yields the familiar 85% world-class OEE figure traced to Seiichi Nakajima's Total Productive Maintenance work. It is a reference point, not an audited requirement, and it assumes an honest ideal cycle time, a 95% performance means little if the denominator was soft. Treat it as orientation; your own measured trend against a fixed ideal is what actually drives decisions, as what counts as a good OEE score makes clear.
The metric is also formally standardized. ISO 22400-2:2014 the international standard for manufacturing operations KPIs defines the effectiveness (performance) ratio and its inputs, ideal cycle time, produced quantity, and operating time, among its 34 indicators, giving the calculation a consistent, auditable basis. Anchoring your performance rate to that definition keeps it comparable across lines and sites, which is the entire reason to compute it the same way every time.
How does the performance rate fit into OEE?
Performance is one of three factors that multiply into OEE, and it is the one that turns speed into a number. Availability comes from the availability rate quality from good-unit yield, and performance from this calculation; multiply the three and you have OEE. Because they multiply, a soft ideal cycle time inflates not just performance but the whole OEE score, which is why the ideal is the input most worth getting right. It is also why per-shift OEE comparisons only hold up when every shift is graded against the same fixed ideal.
All of this rests on measuring total count and run time from the equipment rather than estimating them. A performance rate built on hand-tallied counts and remembered downtime is only as good as the memory behind it, and the ideal cycle time can only be derived honestly if you have real cycle-level data showing the fastest sustained rate. Harmony captures counts and run time from machine signals and preserves the cycle history the ideal is drawn from (see the platform or the CLS results). From there, roll performance into the OEE calculation check it against your plant KPIs and test scenarios in the OEE calculator.