Most SME manufacturers have heard of OEE. Fewer calculate it, and fewer still calculate it the same way twice. This guide walks through an OEE calculation with realistic numbers from a small CNC shop, shows where the usual mistakes creep in, and explains why the quality of your shop floor data matters more than the formula.
OEE, or Overall Equipment Effectiveness, answers one question: of the time you planned to make good parts on a machine, how much did you actually spend making good parts? It is the product of three factors.
OEE = Availability x Performance x Quality
Each factor isolates a different kind of loss. Availability captures stops. Performance captures running slower than you should. Quality captures parts you made but cannot ship. Multiply them and you get one number that shows how much capacity you are really using.
The OEE calculation, step by step
Take a typical example. A precision engineering firm runs a 5-axis machining centre across two 8-hour shifts. Each shift has a 30-minute break. On the day in question the machine made an aluminium bracket with an ideal cycle time of 2 minutes per part.
Step 1: Planned production time
Start with shift time and remove planned stops only. Two 8-hour shifts is 960 minutes. Take off two 30-minute breaks and you have 900 minutes of planned production time.
This is the baseline for everything else. If the machine was never scheduled to run, that time does not count against OEE.
Step 2: Availability
During the day the machine lost 90 minutes to two changeovers and 45 minutes to a spindle fault. That is 135 minutes of downtime, leaving 765 minutes of run time.
Availability = run time / planned production time = 765 / 900 = 85.0%
Step 3: Performance
The machine produced 330 parts. At an ideal cycle time of 2 minutes, those 330 parts should have taken 660 minutes. They took 765.
Performance = (ideal cycle time x total count) / run time = (2 x 330) / 765 = 86.3%
The missing 105 minutes is the hidden loss: short stops, waiting for material, an operator away from the machine for a few minutes at a time, a feed rate turned down on a tricky feature. None of it gets written down, but it adds up.
Step 4: Quality
Of the 330 parts, 16 failed inspection. That leaves 314 good parts.
Quality = good count / total count = 314 / 330 = 95.2%
Step 5: OEE
OEE = 0.850 x 0.863 x 0.952 = 69.8%
You can check it another way. 314 good parts at 2 minutes each is 628 minutes of truly productive time. Divide by 900 planned minutes and you get 69.8% again. If both methods agree, your maths is right.
Worked example summary
| Measure | Figure | Calculation |
|---|---|---|
| Shift time | 960 min | 2 shifts x 480 min |
| Planned stops (breaks) | 60 min | 2 x 30 min |
| Planned production time | 900 min | 960 – 60 |
| Downtime (changeovers and fault) | 135 min | 90 + 45 |
| Run time | 765 min | 900 – 135 |
| Total parts / good parts | 330 / 314 | 16 rejects |
| Availability | 85.0% | 765 / 900 |
| Performance | 86.3% | (2 x 330) / 765 |
| Quality | 95.2% | 314 / 330 |
| OEE | 69.8% | 0.850 x 0.863 x 0.952 |
Notice that none of the three factors looks bad on its own. 85%, 86% and 95% would each pass a quick glance in a production meeting. Multiplied together, nearly a third of the machine’s planned capacity has gone.
What is a good OEE score for an SME?
You will often see 85% quoted as world class. It is a useful reference, but it comes from high-volume, repetitive production. A jobbing shop running dozens of different part numbers a week will spend more time on changeovers and first-off inspection, and its OEE will be lower for sound reasons.
For most SME discrete manufacturers the more useful comparison is with yourself. Measure a machine consistently for a month, get a baseline, then track the trend. Moving a bottleneck machine from 60% to 70% is worth far more than arguing about whether 70% is good.
It also helps to put a value on each point. If a machine’s planned time is 900 minutes a day, one OEE point is 9 minutes of good output. Across a year of working days that becomes a meaningful number of extra parts, from a machine you already own.
Common OEE calculation mistakes
Using the wrong ideal cycle time
Performance depends entirely on the ideal cycle time. Use a padded quoted time and performance looks brilliant. Use a time nobody could ever hit and it looks terrible. Use the best realistic cycle time the machine and process can sustain for that part, and keep it consistent.
Hiding downtime inside planned stops
It is tempting to call changeovers planned and remove them from the baseline. Do that and availability rises overnight without anything improving. Changeovers are a loss you can reduce, so they belong in downtime.
Averaging OEE across different machines
A simple average of a machine at 80% and one at 40% gives 60%, whatever the difference in hours each one ran. Weight by planned production time, or better, report machines individually, especially your bottleneck.
Counting reworked parts as good
Quality should count parts that were right first time. A part that went back for rework consumed extra capacity. Counting it as good hides a real problem.
Calculating it from memory
This is the biggest one. If OEE is worked out on Friday from job cards, operator recollection and a whiteboard, the figures will be rounded, incomplete and optimistic. Short stops are the first thing to disappear, which is why performance is so often overstated on paper.
How to improve OEE in an SME factory
The formula tells you which factor to work on. The fix depends on which one is dragging the number down.
If availability is low
Look at changeovers and breakdowns first. Record the reason for every stop, not just the duration. Once you can see that one machine loses four hours a week waiting for tooling or first-off approval, the fix is often organisational rather than technical. Our article on the hidden cost of machine downtime covers how to put a pound figure on these stops.
If performance is low
Performance losses are hard to see without live data because they happen in small pieces. Common causes in SME shops include operators running two machines and leaving one idle, material not at the machine when a job starts, and programs running conservatively because nobody has reviewed them. Capturing start and stop times on the job itself reveals the gaps.
If quality is low
Track scrap and rework by part, operation and machine. Patterns show up quickly: one fixture, one material batch, one program revision. Fixing the root cause lifts quality and, because rework consumes machine time, often availability too.
Why live shop floor data makes OEE useful
OEE calculated once a month from paper records is a history lesson. By the time you see it, the lost time is gone and nobody remembers why.
When operators clock on and off jobs at the machine and log stop reasons as they happen, the three factors calculate themselves. Run time, part counts and rejects come from the same records used for job tracking and costing, so there is no separate data collection exercise. Supervisors can see today that a machine has been down for 40 minutes, rather than finding out next week.
That is the approach DynamxMFG takes. Shop floor tracking captures job progress and downtime reasons in real time, and the same data feeds dashboards, job costing and scheduling. Typical client results include a 10% reduction in downtime and a 15% increase in efficiency, and Gloucestershire Machining Centre increased capacity by 40% once it could see machine time and profitability by component.
OEE also sits well alongside the other measures that matter to an operations director. If you are building a wider dashboard, our guide to 6 manufacturing KPIs for operations management is a good place to start.
A simple way to start this month
Pick your bottleneck machine. Agree the planned production time, the ideal cycle time for its main parts and a short list of stop reasons. Record every stop, every part and every reject for four weeks, then run the OEE calculation above each week.
You will almost certainly find that the biggest loss is not the one everyone assumed. That is the point of measuring it.
OEE calculation FAQs
What is the formula for OEE?
OEE = Availability x Performance x Quality. Availability is run time divided by planned production time. Performance is ideal cycle time multiplied by total count, divided by run time. Quality is good count divided by total count.
Can I calculate OEE without separate factors?
Yes. Multiply the good part count by the ideal cycle time and divide by planned production time. It gives the same answer, but you lose the breakdown that tells you where to act.
Should changeovers count as downtime?
Yes, in most cases. Changeovers are a loss you can reduce. Treating them as planned stops inflates availability and hides one of the biggest improvement opportunities in a high-mix shop.
How often should an SME measure OEE?
Daily or by shift on your bottleneck machines, reviewed weekly. Monthly figures are too late to act on and too easy to reconstruct from memory.
Is OEE worth tracking in a high-mix, low-volume shop?
Yes, as long as you compare each machine against its own baseline rather than a world-class benchmark. The value is in seeing which losses are growing or shrinking over time.
Want OEE that calculates itself from live shop floor data? DynamxMFG gives UK SME manufacturers real-time job tracking, downtime reasons and machine performance in one system, with a typical go-live inside 90 days.
Keep reading
A 30 minute call, a live look at DynamxMFG, then we configure it on your data and show it back to you.
