I used to think that the most efficient factory was the one whose machines ran faster than everyone else’s. But the story of the “SMED” method drew my attention to something I hadn’t considered: the time a machine is stopped may matter more than the time it is running.

Behind this English abbreviation, which means “single-minute exchange of die”, is a Japanese story that began in postwar factories and ended up changing how factories around the world think about a seemingly simple economic question: How many units should we produce at a time?

What does a die change mean?

In car factories, doors, roofs, and body parts are made using giant presses that stamp metal sheets into heavy dies, with a different die for each part. When the factory switches from one part to another, the press stops, the old die is removed, and the new one is installed and adjusted. It is then tested until the first good part comes off the line. Throughout this period, known as “changeover time,” the machine produces nothing, while costs continue to accrue.

Why did factories produce in huge quantities?

If changing a die takes hours, the logical decision is to produce as much as possible before changing it, so that the cost of downtime is spread across more units.

This logic has a basis in economic theory. In 1913, American engineer Ford W. Harris published an article titled “How Many Parts to Make at Once,” in which he explained that batch size is governed by two opposing forces: the cost of capital tied up in production, which favors smaller batches, and the cost of setup each time production starts again, which favors larger ones. This idea gave rise to the “economic order quantity” formula, which remained the gold standard for determining batch sizes through most of the twentieth century.

The problem, as I see it, is that the formula treated setup costs as a given—as something that could not be changed. The result was factories full of inventory: sprawling warehouses, money tied up in parts waiting their turn, defects that might not be discovered until thousands of units had been produced, and a slow response to any change in demand.

Toyota’s dilemma

After World War II, Toyota was operating in a limited Japanese market that called for multiple models in small quantities. It did not have the capital or the space to stockpile inventory the way giant American factories did. On top of that, die changes took a long time—between two and eight hours, according to some sources.

Taiichi Ohno, who is credited with building the Toyota Production System, was working on the idea of “just-in-time production”: the factory should produce only what is requested, in the quantity requested, and when it is needed. But this idea could not survive in a factory where every changeover took hours, because it would then be forced to return to large batches.

Shigeo Shingo and the distinction that changed everything

This is where Shigeo Shingo (1909–1990) comes in, the Japanese industrial engineer who began teaching courses to Toyota employees in the mid-1950s. Drawing on his experiences in Japanese factories during the 1950s, he arrived at the distinction that would underpin the entire method:

Internal setup: tasks that can only be done when the machine is stopped, such as installing a new die.

External setup: tasks that can be done while the machine is running, such as bringing the new die next to the press and getting the tools ready before stopping. The idea seems obvious once stated, but factories did not distinguish between the two types: first the machine stopped, then people began looking for tools, moving the die, and preparing it. In other words, the machine was waiting for tasks that could have been done while it was running.

One of the best-known stories about the method was told by Shingo himself, about his work at Toyota starting in 1969: a 1,000-ton press took about four hours to change over. That time was reduced to an hour and a half, after which management asked them to bring it down to three minutes. According to Shingo, the key to this leap was converting as much internal setup as possible into external setup, and the target was achieved within a few months. That is where the name came from; it does not literally mean one minute, but rather a time expressed in a single digit—that is, less than ten minutes.

Another account from inside Toyota

To be fair, this story is disputed. In an interview published by the Lean Enterprise Institute in 2006, Isao Kato, a former Toyota training and production executive, said the press story was told in an inaccurate sequence, and that reducing changeover times was a collective effort that began in the late 1940s. The average had fallen to 15 minutes by 1962, and then to three minutes in the press department by 1971, thanks to the company’s engineers. At the same time, however, he credited Shingo with the distinction between internal and external setup.

I tend to think that the disagreement over who deserves the credit does not diminish the value of the idea. Toyota achieved the results on the factory floor, and Shingo turned them into a written method that could be shared in his book “A Revolution in Manufacturing: The SMED System,” published in English in 1985. The idea then spread from Toyota’s factories to factories around the world.

How does the method work?

The Kaizen Institute sums up the method as a series of steps: film the changeover as it actually happens and measure each task; then separate internal tasks from external ones; convert as many internal tasks as possible into external ones—for example, by preheating dies and standardizing their height to avoid readjustment; then simplify the remaining tasks by using quick-release fasteners instead of bolts and having two workers operate at the same time; and finally, organize tools near where they are used.

Shingo is often quoted for a witty observation: only the final turn of a screw actually tightens it; all the turns before that are just movement. In a 2022 article published by the Lean Enterprise Institute, Art Byrne says that simply separating internal and external tasks usually reduces changeover time by 50% or more, at no cost.

A close analogy for all this is a Formula 1 pit stop, which the Kaizen Institute draws on to illustrate how tasks can be distributed and carried out simultaneously. There, all four tires are changed in a matter of seconds; the Guinness World Record of 1.80 seconds was set by McLaren at the 2023 Qatar Grand Prix. The secret is not just fast hands, but the fact that everything that can be prepared before the car arrives has already been prepared.

Why did SMED turn the logic of mass production on its head?

This is where the economics really come into play. Harris’s formula was not wrong, but it treated setup costs as fixed. Toyota refused to accept that assumption. Instead of asking, “How many units should I produce to spread the cost of a changeover?” it asked, “How can I reduce the cost of the changeover itself?”

When changeover time falls from hours to minutes, the economic batch size falls with it, because the reason for producing large batches has disappeared. The Kaizen Institute notes that Ohno’s ultimate goal was “zero changeover,” where the batch size approaches a single unit.

With smaller batches, inventory and the money tied up in it shrink, and less space is needed. Byrne estimates that large batches require 40% to 50% more space. Defects are detected early, before they recur across thousands of units; the factory can respond quickly to changes in demand; and every minute saved becomes additional production capacity without buying a new machine. In one example Byrne recounts, the American company Wiremold reduced a changeover on one of its rolling machines from 840 minutes—that is, 14 hours—to six minutes. According to Byrne, the company’s efforts to reduce changeover times helped shorten lead times from weeks to a day or two, with little or no capital expenditure.

And so the meaning of efficiency itself changed. SMED did not abolish economies of scale, but it stripped them of their monopoly on efficiency, shifting the advantage from the factory that produces more to the factory that can switch more quickly between what the market demands.

And what does this mean for us?

In the Kingdom, where the Future Factories Program aims to transform 4,000 factories into smart facilities, discussion usually focuses on automation, robots, and data—and rightly so. But the SMED story reminds us that some of the biggest gains do not require a new machine, just a new way of looking at how work is done. In my view, one of its best features is that it is accessible to everyone: a small factory can take the first step with a phone camera, a stopwatch, and a team willing to observe.

When I think again about that press that used to sit idle for hours, I realize the problem was not the weight of the die, but the assumption surrounding it: that downtime was unavoidable, and that the only solution was to produce more to make up for it.

That is why perhaps the best way to describe the SMED method is not as a way to speed up die changes.

A more accurate description, as I see it, is a way to rethink downtime itself: what must happen while the machine is stopped, and what can happen while it is running. When the answers to these two questions change, so do batch size, inventory levels, and the factory’s ability to produce what the market demands, rather than produce first and then wait for someone to buy it.

That is why I think the SMED story is worth reading from an economic perspective—not merely as an engineering trick in car factories, or as a chapter in Toyota’s history, but as a window onto a broader shift in the logic of production:

From large batches that compensate for slow changeovers, to fast changeovers that make small batches viable.

Disclaimer: This article is for educational purposes and aims to explain the subject in a clear, accessible style. The information it contains is drawn from published sources and has been adapted and simplified; it does not represent an official position or statement by any organization. Some details of this historical story vary across accounts and sources, so readers are encouraged to consult them for the full picture.