OEE Explained: The One Factory Number Worth Tracking

What is OEE in simple words?

OEE is a way of asking one question about a machine: of the time we planned to produce, how much was truly productive? It was developed in lean manufacturing and is now used across industries from packaging to machining. Instead of looking at how many parts came out, it looks at the gap between what was possible and what actually happened.

The gap is split into three kinds of loss. Time lost to stops is an availability problem. Running slower than the machine's ideal speed is a performance problem. Making parts that must be scrapped or reworked is a quality problem. Splitting the gap this way is what makes OEE more useful than a single output count.

How do availability, performance and quality work?

Availability compares the time the machine actually ran with the time it was planned to run. Breakdowns, waiting for material, setup and changeovers all reduce it. Performance compares the actual speed with the ideal speed, so small stops, slow cycles and operators running below rated speed all reduce it. Quality compares good parts with total parts started, so scrap and rework reduce it.

These three are expressed as fractions and multiplied together. If each one is a little below perfect, the combined figure drops faster than people expect, which is exactly why a machine that looks busy can still be delivering far less than its potential.

• Availability: running time divided by planned production time
• Performance: actual output divided by what ideal speed would have produced in the running time
• Quality: good parts divided by all parts made
• OEE: availability multiplied by performance multiplied by quality

What does a worked example look like?

This example uses made-up numbers to show the method. Suppose a machine is planned to run 8 hours in a shift. It stops for 1 hour because of a breakdown and a changeover, so it runs 7 hours. Availability is therefore 7 out of 8.

At its ideal speed the machine could make 100 parts an hour, so 700 parts in 7 hours. It actually made 630. Performance is 630 out of 700, which is 9 out of 10. Of the 630 parts, 603 passed inspection, so quality is 603 out of 630, a little over 19 out of 20 (roughly 0.957 as a decimal).

Multiply the three: 7 over 8, times 9 over 10, times about 0.957 gives roughly 0.75, or about three out of every four planned hours of ideal good production. The remaining quarter is lost across stops, slow running and rejects, and the breakdown of those three shows which one to attack first.

Why is OEE better than counting output?

Output totals hide the reasons behind a shortfall. If a shift produces fewer parts than planned, the total does not tell you whether the cause was a long stop, a worn tool slowing cycles, or a batch rejected at inspection. Each cause has a different owner and a different fix, so treating them as one number leads to guessing.

OEE also makes machines comparable. A fast machine with long stops and a slower machine that runs steadily can be judged fairly. Be careful though: comparing OEE across very different processes is rarely useful. The most valuable comparison is a machine against its own history.

How do you collect the data for OEE?

Many plants start with paper logs filled in by operators. This can work for a pilot, but handwritten data is often incomplete and reconstructed at the end of the shift. Automatic capture is more reliable: machine status signals give run and stop times, and a part counter gives output without relying on memory.

Reasons for stops usually still need a human touch. A simple tablet or button panel where the operator selects a reason from a short list gives better data than free-text notes. Keep the list short, such as no material, breakdown, changeover and quality check, and review it monthly so that the category called other does not become the largest.

• Agree the planned production time for each shift before measuring
• Define what counts as ideal speed, ideally from the machine rating or best proven run
• Record stop reasons from a short, fixed list
• Count rejects and rework separately so quality is not hidden
• Review the data weekly with supervisors, not only management

How should owners use the number once they have it?

Start with the biggest loss category on your bottleneck machine, because an hour recovered there is an hour recovered for the whole line. If availability is the weak point, study changeover and breakdown patterns. If performance is low, look at micro-stops, material feeding and operator practice. If quality is low, look at tooling, settings and incoming material.

Avoid turning OEE into a target to punish. Teams that fear the number start rounding it up. Treat it as a diagnostic, publish it at the machine, and celebrate improvements. A Plus Solution builds dashboards that calculate these figures automatically from machine signals, but the discipline of acting on them matters more than the tool.

Frequently asked questions

What is considered a good OEE?

There is no single answer, because it depends on the process, product mix and how strictly ideal speed is defined. Compare a machine with its own past results first, and use published benchmarks cautiously.

Can small workshops use OEE?

Yes. Even a single bottleneck machine tracked with a simple log gives useful insight. You do not need a full automation project to start learning where time is lost.

Is OEE the same as productivity?

Not exactly. Productivity usually compares output to inputs such as labour or cost. OEE focuses on how effectively a machine uses its planned time to make good parts.

Should planned maintenance count as lost time?

That is a policy choice. Many plants exclude planned maintenance from planned production time, while others include it to see the full picture. Whichever you pick, apply it consistently so trends stay meaningful.