OEE in Lean Manufacturing: A Map of Your Losses, Not a Score

Picture a lean board with OEE at the top. Six months on, the number is up, and so is the stock in the warehouse.

The math is fine. The line made more parts nobody had asked for, and the number rewarded it.

OEE does belong in lean. This page shows the job it does there, where that job stops, and how to use it for kaizen without fooling yourself.

What OEE does for a lean plant

Lean asks whether value flows to the customer with as little waste as possible. Our page on lean manufacturing principles covers that bigger picture.

OEE (overall equipment effectiveness) asks a smaller question: of the time this machine was meant to run, how much became good product at full speed?

The Lean Enterprise Institute calls it "A total productive maintenance (TPM) measure of how effectively equipment is being used." So it looks at a machine or a line, not at the whole value stream.

Inside that limit, it does three jobs for a lean team:

• Seeing: losses that feel normal on a shift get a size.
• Choosing: the biggest named loss becomes the next kaizen target.
• Checking: the same count, before and after, shows whether a fix held.

Read each OEE loss as a kind of waste

The Institute defines muda as "Any activity that consumes resources without creating value for the customer." Every OEE loss fits that line.

Here is how the six big losses map to the seven wastes. The mapping is our reading, not a standard.

Lean also sorts waste in a way that helps you set targets.

The Institute's type one covers "activities that cannot be eliminated immediately". Type two covers those that "can be eliminated quickly through kaizen".

In our reading, a changeover on today's tooling is usually type one, so you shrink it first. A jam with a known cause may be type two, so you remove it this week.

The waste OEE cannot see

Overproduction, transport, unnecessary processing and stock sit outside OEE. A forklift moving pallets around the plant costs the line no OEE at all.

Overproduction is the one to worry about. The Institute notes that Ohno considered it "the most grievous form of waste because it generates and hides other wastes".

For the two lists side by side, read six big losses vs seven wastes.

Worked example: one shift, in minutes

All numbers here are example data, for one packing line with the same ideal rate on every product. The full method is in our OEE calculation guide.

• Availability: 390 / 450 = 86.7%
• Performance: 33,150 / (390 x 100) = 85.0%
• Quality: 32,487 / 33,150 = 98.0%
• OEE: 86.7% x 85.0% x 98.0% = 72.2%. As a check, 32,487 / (450 x 100) gives the same.

A lean team cannot do much with 72.2%. Turn each loss into minutes at the ideal rate, and it can.

Look at the speed loss. It is 58.5 minutes, nearly as much as every stop combined, and no one wrote any of it down.

It is the jam cleared in 20 seconds, and the rate someone turned down after a bad week. Stops under your short-stop limit count as speed loss, not as stops.

See micro-stops for how to find them.

Now the team has three targets with a size on each, instead of one percentage.

From a lost minute to a kaizen target

A good kaizen target has four parts. OEE data gives you the first two.

1. A named loss with a size. For example: changeover, 40 minutes a shift.
2. A pattern. Which product change, which shift.
3. A reason it matters. The line is your constraint, or it holds up a delivery. If not, the gain may only become stock. Theory of constraints vs lean goes deeper on this.
4. A check, agreed before you start. Same line, same products, same way of counting.

Take the changeover. The Institute says SMED (single minute exchange of die) aims for "a single digit, or less than 10 minutes".

A first step is fine, so say you cut it from 40 to 25 minutes.

• Run time goes from 390 to 405 minutes.
• Availability goes from 86.7% to 90.0%.
• OEE goes from 72.2% to 75.0%, up 2.8 points. That is about 1,250 more good parts a shift, in this example.

To plan and run the event itself, read how to use OEE data in a kaizen event.

Whether those 1,250 parts are worth anything is a separate question. That is where the score starts to mislead.

Why OEE misleads when it becomes the score

Here are three moves on the same example line. Each one changes the number in a way a lean manager would argue with.

Move 1: stop when demand is met

Say demand is 28,000 good parts a shift, and the line can make 32,487. Run to the end of the shift, and OEE stays at 72.2% while 4,487 parts go to stock.

Stop at 28,000 instead, and with the same 60 minutes of stops you get there after about 396 minutes. If the rest stays in planned time, OEE is 28,000 / 45,000 = 62.2%.

Toyota describes just-in-time as "Making only what is needed, when it is needed, and in the amount needed". So the plant did the right thing, and the score fell 10 points.

The fix is a definition. Take time with no demand out of planned time, under its own reason, and OEE is 28,000 / 39,600 = 70.7%, not 62.2%.

Only real no-demand time qualifies. Moving other stops out just flatters the number.

Then watch that free time separately, with utilisation or TEEP.

Move 2: half the batch, twice the changeovers

Cut the changeover from 40 to 20 minutes, then change over twice a shift. Total changeover time is still 40 minutes, so OEE does not move.

Yet batches are half the size, so less stock waits between steps and lead time can shorten. Lean would call that the point of the exercise, and a plant watching only OEE would see no result.

Move 3: shorter changeover, same batch

This is the kaizen from the last section, and the one the score likes best. It pays in parts only if someone needs them.

The same goes for speeding up a machine that is not the constraint. OEE rises, and the extra parts wait in front of the next step.

Other ways the number misleads

• Ranking lines against each other. Take two identical lines, one with a single product and one with ten. Vorne's OEE.com (read 9 November 2026) says the second line "is likely to experience far greater Availability Loss due to Changeovers".
• A plant average. One line improving can hide another getting worse, and no one can act on an average.
• Tying it to pay. People protect the number, so ideal rates soften and stops get new labels. See how OEE gets gamed.

Use OEE to ask where the time went. Do not use it to ask who is to blame.

Where TPM comes in

OEE came out of total productive maintenance. OEE.com credits Seiichi Nakajima, who led its introduction in the early 1970s at the Japanese Institute of Plant Maintenance.

TPM gave lean plants two ideas that still matter on the production side.

• Losses belong to everyone. The Institute says TPM "requires the total participation of all employees", including line managers, engineers, quality staff and operators.
• Name the losses and work through them. That habit is the same one a good kaizen routine needs.

You do not need a full TPM programme before you measure OEE. You do need a regular time when those people look at the same named losses.

Our page on OEE and TPM tells the full story.

Lean metrics to keep next to OEE

OEE needs company. A short set of lean manufacturing metrics, or lean manufacturing KPIs, covers what it cannot see.

Takt anchors this list. Its purpose, says the Institute, is "to precisely match production with demand".

The value is in reading them in pairs:

• OEE up, lead time down: a real win.
• OEE up, stock up: you are probably overproducing.
• OEE flat, batches smaller: also a win, so say so on the board.
• OEE down, stock down, delivery steady: check your planned time before anyone panics.

For the wider plant view, see manufacturing KPIs.

Try this on one of your own lines

Pick one line and last week. Then try to answer these without opening a spreadsheet.

1. How many minutes did it lose, and to which three reasons?
2. How much of that loss has no reason at all, because the stops were too short to write down?
3. Does the biggest loss sit on one product change or one shift?
4. Did your last kaizen shrink the loss it was aimed at, on the same count?
5. Did the extra output go to a customer, or to the warehouse?

Questions 1 to 4 need machine data and stop reasons. Question 5 needs your ERP and your lean board.

If the honest answer to the first two is "roughly", your kaizen list is being set by whoever remembers the shift best.

How Fabrico fits

Fabrico is a cloud-based OEE and light MES platform. It covers the machine end of this page: OEE, stops and reasons by line and shift.

• Stops: they come from the machine signal, through a PLC or a Fabrico IoT device, including stops too short to log by hand.
• Reasons: the operator adds the reason on a tablet or phone. That is what turns lost minutes into a kaizen list.
• Parts: AI cameras (an add-on) count parts.
• Video: when you review a stop, you can open the camera video for that minute. In the standard setup it is streamed, not stored in Fabrico.
• Maintenance: built in on the same machine data, as a complement.

Lead time, stock and order routing stay in your ERP and on your lean board, and production planning and scheduling are not part of Fabrico today. So it feeds your lean system, and does not replace it.

Ficosota (also a Fabrico investor) raised OEE from 53.1% to 59.6% on 16 lines, June to September 2026, with monthly good output up 23.2%.

Connecting the machines takes about one month, and rollout across the lines you choose takes 3 to 4 months. From then on, the answers to questions 1 and 2 come from your own lines.

See how the pieces fit in our MES and OEE overview. If cameras fit your line, ask us to show them in the demo.

Want to see lost minutes and reasons on lines like yours? Book a demo or Contact us.

What people ask next

Is OEE a lean tool?

Not by origin. It comes from TPM, and lean plants took it up because it puts a size on equipment losses.

Treat it as a measure. It tells you where to look, and tools such as SMED do the fixing.

What is a good OEE in a lean plant?

There is no lean target. OEE.com (read 9 November 2026) says "There is no one ideal OEE score", and notes that the 85% often quoted has roots in 1970s Japanese car making.

The better test is whether your biggest named loss is shrinking. For published figures, see our OEE benchmark page.

Should a machine that is faster than takt run flat out?

No. If it runs ahead of demand, it makes stock, so let it stop and use the free time for training or checks.

Book only true no-demand time under its own reason, outside planned time. Changeovers stay inside planned time.

Can OEE go down while lean is working?

Yes. It happens when no-demand time stays inside planned time, or when more frequent changeovers cost more minutes than you saved on each one.

Check lead time and stock before you read that as a failure.

Can I start OEE for lean on paper?

Yes, and it teaches a team the loss language. The catch is that short stops rarely make it onto a sheet, so speed loss stays a mystery.

Our comparison of light MES vs spreadsheets shows when to move on.

Do I also need lean software?

That is a different job. OEE data tells you where the time went, while 5S audits and kaizen boards manage the work that follows.

Our list of lean manufacturing software covers those tools.

Curious where the minutes go on lines like yours? Book a demo or Contact us.