Lean Conformance vs. TPS Performance

Art Smalley
Art Smalley

Here is a response to a question about multi-purpose cells I gave to a question over on the Lean Edge.

The main question asked here is “have workplaces moved to multi-purpose cells or do we still see isolated operators on the shop floor (現場 / Genba)”? The statement implies that was what “Toyota” was teaching us 20 years ago.  Well that last part I sort of doubt it. In reality that is partly what the observer was learning or partly what the instructor was relating at the time. Unfortunately that is not the right frame for implementing TPS with success. I will probably not answer the question in the way the person posing this might have expected so apologies up front. The “problem” with this question is that it falls into the typical lean failure mode of looking for implementation of some tool or rule or method or other technique. Do we see enough “conformance” to some Lean stereotype. Do you see enough 5S? Do you see enough Standardized Work? Do you see enough multi-purpose cells or value stream maps, etc. This is the unfortunate reality of Lean in most companies or training sessions I observe. In contrast TPS historically is different. I will attempt to explain why and give some simple points of background reality about the history of multi-process cells in Toyota.

The lean movement sadly is still mostly about “tools” or “rules” or “methods” as I have explained in the past. The game being played out in most instances is to copy the tool, or follow the advice of some sage sensei and implement that tool or rule. It is about “conforming” to some notion of lean…the entire point of “performing” seems to get lost. 5S, Value Stream Maps, Standardized Work, SMED, etc. etc. comes to mind. Lost in all of this the point of “why” something is being done and “results” i.e. the need to improve in some way to improve customer value or solve some fundamental problem like in productivity, quality, delivery, etc.

I’ll try to explain the problem with this approach by way of a golf analogy I often use. The Lean Golf Movement aka LGM (Ok I just made that up of course) states that a proper Lean golf swing consists of a pre-shout routine, a stable stance, a flowing swing, a swing in rhythm, and solidly striking the ball. Those are the five rules and methods, etc. Are we seeing enough of a flowing swing? That is what the Lean Golf Movement cares about.

The answer to that question is I dunno. And I don’t really care. I am a TPS person. I want to know what you shot on your last full round of golf? No bullshit. Par for 18 holes is 72 on most courses…the average golfer shoots 90 or worse most of the time. Kind of like the average company is a bogey performer as well…In order to move from 90 to 72 you have to improve 18 shots on average…where exactly are you losing them? Putting, Short game, Accuracy of greens or fairways hit in regulation? Combination of all the above? How much in each case? Why? Let’s figure out the actual problems with facts and data and get started fixing what is not performing well enough and see if we can bring down that 18 shot problem (gap from standard). Don’t expect a miracle over night.

The Lean Golf Movement mainly cares more about how it looks in its new clothes and how it walks down the fairway. And of course if the swing is “flowing” enough. It believes in the 14 ways of the Lean Swing, Respect for the Game, and of course Lead-time around the golf course. They draw maps of where the ball goes. However the Lean Golf Movement does not like to talk about its actual “score”. LGM asks if our 7 iron shot flowed nicely in a multi-process swing approach. That in a nutshell is the big difference I see between Lean and TPS whether it is my golf analogy or actual companies attempting Lean. I don’t care about verbal descriptions of your golf swing. As the pro’s joke there is no room on the score card for descriptions of how you hit the ball. After each hole in golf you only write down what you actually scored. No bullshit.

So what did Toyota “score” with multi-process cells? In 1950 or so when Mr. Ohno started to embarked upon improvements in the machine shops of Toyota one big problem (there were many keep in mind) was that the average productivity of Toyota was only about 1/9th of that compared to U.S. car manufacturers. That is a big problem for a company trying to improve and compete. In historical context you have to remember that volumes were exceedingly low during this time so “takt time” was long as in several minutes or more in some cases. I can’t find the exact quote off hand from Mr. Ohno but it went something like this…”I did not think that meant that workers in other countries were working nine times harder than we were. Rather it was expressing that our own style of work was inefficient and full of waste. We needed to change our way of production and eliminate waste to become more productive”.

You have to understand in hindsight also that processes were highly manual in those days as well. So a good way to bring up productivity is to have workers complete more than one task and not simply overproduce the wrong items. So the human work was re-organized to follow the product flow and the work speed was paced to takt time. Instead of running one type of mill all day and overproducing you’d run the mill, then the drill, then the lathe, then go back to the mill and start all over again every few minutes (i.e. to takt time). Mr. Ohno observed some of this style of production in the loom business of Toyota where he originally worked. He wanted to carry it over to the engine plant where he was now manager in order to improve productivity among other things.

Of course I am over simplifying all the mini-details but it is key to realize that this was a step in a process of solving a specific problem – that of 1/9th the worker productivity level of the United States in car manufacturing. Of course this also involved eliminating waste from the worker’s job sequence, simplifying and standardizing things, and building in quality the first time. Some of this we now call “multi-process cells” in English. As a result Toyota applied these concepts and many others as well and gradually closed the productivity gap.

Toyota slowly evolved from one person running one machine all day to multi-process handling. This became one man two machines, one man three machines, one man four machines, etc. etc. over time. When Shigeo Shingo first visited Toyota in 1955 he wrote in one of his early books in Japanese the following passage. “I first visited Toyota in late 1955 to start teaching the P-Courses. I was shocked at the extent of multi-process handling I observed. In one extreme case I saw one person operating as many as 17 processes and the average was five machines.” Mr. Ohno’s main engine manufacturing lines had gone pretty far in a few cases even by late 1955. Again keep in mind this is due to the fact that takt times were long in those days, productivity was low / waste high, and processes were extremely manual back then. The “result” of all this steady work was that by the early 1970’s twenty years later Toyota was more productive than car manufacturers in North America. Quality was also better in many cases as well…

So in Toyota’s historic case and this specific type of problem implementing multi-process handling with one person completing multiple tasks in accordance with takt time worked for improving worker productivity. And yes it should work in a lot of cases. However they key part is solving a fundamental need to improve or problem and obtaining actual results (i.e. performing rather than just conforming). Multi-process handling is pointless if it gains nothing. Lean often confuses the “ends” versus the “means” in this way. It gets hung up on are we flowing enough, or do we have our value stream maps drawn right, or do we have our multi-process cells in place? Do we look right? It neglects to define the problem, identify the gap and goal, identify the root causes or ways to improve, and measure before and after to see if there is actual improvement. TPS keeps detailed score. Lean jumps to “implementation” of a tool, rule, method, or concept without adequately understanding why or what they will improve. The result is usually not much improvement.

The other odd part of this question revolves around the notion of “cells” as part of the universal solution space for conforming to for Lean. I think it would be an excellent assignment to visit an actual Toyota Engine Plant and to count the number of U-shaped work cells you can find today. The answer is not many if any…you’ll mostly find long straight highly automated lines which are well organized. The same is true for an engine assembly line. The worker productivity will be high and they’ll be glad to tell you their score for performance improvement over time. No Bullshit.