Six Sigma and beyond: Product Design Opportunities
by David R. Schwinn

Welcome back to "Six Sigma and Beyond." After last month's column, one of our users pointed out errors in Six Sigma, the book I referred to last time by Mikel Harry and Richard Schroeder (New York: Doubleday, 2000). The specific errors in question are in the Six Sigma Conversion Table near the end of the book. There are, in fact, several errors throughout the table. If you need to know defects per million (dfm) at a specific sigma value, you may want to go to a different reference. Most of the errors seem to be at sigma values below 1.55, so the table may be fine above sigma values of 1.50, but I was unable to verify that. I spot-checked the table against a standard normal table in Total Quality Tools (PQ Systems, Inc., 1994) at sigma values of 2.0 and 2.5 and they seemed pretty close.

So thanks, Ken, for the heads up! Ken's assistance got me to thinking about a lesson we learned about performance beyond 3 sigma while I was still at Ford Motor Company, and Ford was trying to develop a corporate culture of continuing improvement. Shortly after W. Edwards Deming came to Ford, he began talking about continuing improvement. At that time, just meeting the specifications made us very happy campers. Since we were having a difficult time doing that, we could see no value in being any better.

About that time a very interesting story came to us from the Transmission Division. It seemed that Ford's new plant in Batavia, Ohio was having capacity limitations in manufacturing a new front wheel drive transmission. In order to meet customer demand, Ford had asked Mazda in Japan to make the same transmission for us, using the same prints and specifications. Once the transmissions got into customers' hands, there were a few customer complaints, just as one would expect from any new transmission. But those complaints were limited to the Batavia-built transmissions. Although there was no way for dealers or customers to know the difference, the Mazda-built transmissions got no complaints!

As you might expect, an investigation ensued. It found that the time-to-shift parameter, the source of most of the customer complaints, was just barely outside the specs for the Batavia-built transmission, while the Mazda-built transmission was well within the specs. It was near six sigma. Beyond convincing Ford about Deming's demand for continuing improvement, the investigation revealed something else that was a total surprise; the Mazda-built transmission was centered in the upper quarter of the specification!

While Ford had developed the specifications based on a compromise between what the design engineers thought customers would tolerate and the manufacturing engineers thought they could produce, the Mazda engineers instead asked the customers what would delight them. They aimed at that target and got as close to it as they could, subject to cost constraints and Ford's specifications. Of course, in examining the key transmission components, the variation in their dimensions, finishes, and cleanliness all exhibited the same kind of characteristics as customers had seen in their transmission performance.

The moral of that story is that Six Sigma not only permits, but--if it is to be effective--demands an entirely new approach to product design. Customer input should be included in such a way that it provides a performance target or targets. Determining only the band of acceptable performance no longer cuts it.

For more discussion of this approach to Six Sigma, take a look at Module 12, "Continuing Improvement Strategies," a video in the original Transformation of American Industry training system. Anyone at PQ Systems can help. If you'd like me to go into more depth or cover any other Six Sigma topic, contact me at support@pqsystems.com. øDavid R. Schwinn