Instead of being showered with thanks, however, Baron was greeted with derision. "They laughed because all their data told them that they were making great doors," says Baron, president of the Center for Automotive Research (CAR), a Michigan-based automotive think tank. "The problem was, they were gathering their data from measurements, not from customers."

Indeed, the engineers had measured and re-measured their door assemblies thousands of times, but didn't know that during everyday use, water was leaking through them. Still, Baron says, those engineers weren't the first, nor the last, to assume that their measurements were the recipe for quality and reliability.

Today, Baron has a message for such engineering teams: If you assume you'll get better quality by demanding excruciatingly tight part tolerances, then you're destined for second-class performance. High reliability, he says, isn't achieved by building perfect components. For proof, he cites Asian automakers. According to surveys done by the Consumers Union and J.D. Power and Associates, Asian automakers have been building the most reliable vehicles for nearly three decades. This year, J.D. Power named Japan's Lexus as the most dependable nameplate, while Consumer Reports magazine cited 46 Asian vehicles among its top 54 "good bets" for used cars. Consumer Reports, which placed no Asian vehicles on its list of 34 "bad bets," said its surveys even revealed that an eight-year-old Lexus had fewer problems per hundred than a six-month-old BMW 7 Series. The reason for that quality dominance, Baron says, is that Japanese automakers make fewer measurements and have more imperfect parts. Whoa, did we say imperfect parts? Indeed, we did. Baron's studies have shown that, incredibly, his studies have shown that stories about Japanese attention to detail on every component are more folklore than reality. The manufacturing methods that have provided such reliability for Asian vehicles have more to do with consistency and good practice than with perfection, say Baron and his colleagues at CAR.

"It's counter-intuitive," notes David Cole, chairman of CAR and a former professor of automotive engineering at the University of Michigan. "It doesn't appear to make rational sense, but if you can build repeatable, imperfect parts, you can solve a lot of quality problems at the system level."
No Micro-Managing

Baron originally learned those counter-intuitive truths while working on his Ph.D. thesis during the late 1980s. His thesis, funded by American automakers and steel companies, took him to manufacturing plants in the U.S., Europe and Asia. During the course of his travels, Baron stumbled on a surprising fact: In Japanese plants, engineers weren't monitoring their processes as closely as their North American and European counterparts.

"We asked, 'How can they be getting high quality when they're hardly ever measuring their parts?'" Baron recalls.

What he and other researchers found was that the Japanese ran stable processes.

"If you know you have a stable process, then you don't have to measure it as often," he says.

Surprisingly, he also found that roughly 20 percent of the parts in those plants were out of "spec." How, he wondered, could system quality be so high when component quality was so low? The answer was hidden in the process, he learned. Instead of focusing on individual part quality, Japanese engineers attended to system quality. By doing so, they could take a conglomeration of imperfect components and assemble them into a reliable system.

Baron compares their process to that of a carpenter nailing baseboard into a sheet of drywall. Instead of cutting several pieces to tight specs, the carpenter could simply lay all the pieces together, then cut the last one to fit. The result could still be a high-quality assembly, Baron says.

"All the carpenter would have to do is make an adjustment to the last board, and then the entire assembly would be perfect," he says.

"There's a misperception about Japanese quality," Baron explains. "People believe that Japanese automakers build all their parts very precisely, and it just isn't true. "We found they don't care about making each part to a precise spec. They just want the assembly to be in spec. And they don't micro-manage the process."

To be sure, none of the proponents of this technique are saying that perfect parts are bad. Rather, they believe that the pursuit of perfection can be elusive at best, and can end up draining too much time and money.

"Personally, I have no problem with perfect parts," Baron says. "But if making perfect parts is prohibitively expensive, then it doesn't make sense."

Expressed in conventional quality terms, Baron's thesis says that the Asian method focuses on the so-called "Cp," a measure of one's ability to produce consistent results from lot to lot. Americans and Europeans, meanwhile, focus on the so-called "Cpk," which tells how well a particular part fits within an absolute measurement value.

"The Japanese want low variation from part to part, and then they deal with out-of-spec measurements as they go along," Baron says.
The Essence of Reliability

Dubbed "functional build," the method has started to gain traction among U.S. automakers.

"General Motors has been using these practices for the last few years and Chrysler has begun to move in that direction," Cole says. "Ford has been a little slower, but they've started using it, too." Still, Baron says that U.S. and European adoption of the technique is spotty, while Asian adoption, particularly by Honda and Toyota, is commonplace.

Automakers were reticent to talk about their use of the process on their manufacturing lines for this article, citing competitive issues. North American tool makers, however, say that the technique offers advantages for them, as well as their customers, the automakers.

"It's the best way to do things, but obviously we can only use it if the customer wants it," notes Mark Schmidt, president of Detroit-area-based Atlas Tool, Inc., which makes stamping dies for automakers.

Schmidt says functional build saves money for die manufacturers and automakers alike. He cites an example of an automaker that employed functional build during the creation of a vehicle roof, during which eight points on the roof were found to be outside the specification. Instead of having to fix all eight, however, the automaker asked Atlas to fix only three. Using functional build, automotive engineers decided that the assembly would fit better, and the dies would be finished sooner and would cost less, if they fixed only the three critical areas.

"If they didn't use functional build, the automaker wouldn't have known which tolerances were important and which weren't," he says. "But because they focused on the quality of the end product, and not just the part, they ended up with a better product for less cost."

Atlas and other manufacturers say that they've used the technique on vehicle roofs, hoods, trunks, doors, cross members, and frame parts. Baron adds that the method can also be applied to plastic and electronic assemblies, such as instrument panels.

Those who use functional build contend that it makes more sense than the traditional Cpk-based methods. Manufacturers who try to make perfect components are trying to do the impossible, they argue.

"A body-side (assembly) can never be 100 percent to spec, at any stage of production," says Schmidt of Atlas Tools. "It just can't be done."

Yet, many domestic automakers continue to push the concept of perfect parts, despite their knowledge that parts made from sheet metal and other materials often get distorted during manufacturing. Worse, compensating for such distortion may cause parts to fall out of spec elsewhere.

Baron recalls one project on which an automaker called for 1,400 check points on a door die. "Most designers think they're defining good quality when they call out 1,400 check points on a mold," Baron says. "But no one can track 1,400 check points. They say they're trying to make a perfect part, but what they're really doing is driving up the price of the tool."
Beyond Conventional Wisdom

Proponents of functional build argue that the concept's proof lies in its success with Asian automakers. By virtually all objective measures, Japanese automakers have exhibited high quality for at least a quarter-century. Consumer Reports, which tears down vehicles and surveys more than 600,000 owners every year, has repeatedly given high marks to vehicles made by Honda, Toyota, Nissan, Mazda and Subaru. No Asian vehicles made Consumer Reports "bad bets" list this year, and some of the lowest-priced Japanese models outclassed luxury vehicles in reliability. The $14,000 Honda Civic, for example, fared better this year than the Audi A6, and the BMW 7 Series, as well as Mercedes-Benz's C-Class, CLK, M-Class and S-Class vehicles.

"In most cases, the most reliable vehicles came from Honda and Toyota," says David Champion, director of testing at Consumer Union's automotive test facility. "And their vehicles are just as complicated, and have just as many electronic features, as a Mercedes or a BMW, yet they seem to get it right."

To be sure, Champion and other experts don't question the first-class performance- of many German vehicles. But the reliability race, they say, is another matter.

"Reliability was what originally put Japanese cars on the map," Champion says. "Their first cars—Corollas, Stanzas, and Accords—were not particularly dynamic. But consumers knew they would work."

Still, North American and European automakers have been slow to adopt their techniques, for reasons ranging from fear of change to corporate politics.

"It's hard to change conventional wisdom," says Cole of CAR. "And conventional wisdom has always said that perfect parts make perfect systems."
Design Experience Needed

Functional build proponents do acknowledge that adoption of the methodology can be very difficult. Today's conventional practices are easier to keep in place, they say, because they don't call on anyone to take risks. Such methods give designers a solid, quantitative reason for rejecting a part.

"Nobody will ever get in trouble for saying that a die is not perfect," Baron says. "It's easier to pretend you're going to make perfect parts, and then make decisions based on that."

Schmidt of Atlas Tools adds that functional build calls for real design expertise. "It takes an experienced designer to say, 'Don't worry about that tolerance, it's meaningless for this part,'" he says. "But an inexperienced designer won't know which tolerances to forgive, so he'll forgive nothing. He needs everything to be perfect."

Schmidt believes that the key to functional build is at atmosphere of cooperation between OEMs and vendors. Both sides, he says, must be willing to work together and forego the finger-pointing that often takes place between the two groups. Moreover, engineering departments must be willing to invest more time and effort in the manufacturing process, with the knowledge that manufacturing costs will drop dramatically as a result. Only then, he says, will engineers be willing to back away from the idea that perfect parts make perfect systems.

"It's not that perfect parts are bad," Cole concludes. "It's just that they're awfully hard to build."