The Wall Street Journal

Rethinking Quality Improvement

September 19, 2005;
Page B3

The love affair between American managers and a wonkish mistress is showing signs of strain.

Managers have been falling head over heels for quality-improvement programs for years. The systems -- which go by unblinkingly nerdish names such as TQM, ISO 9000 and Six Sigma -- differ in details, but they all aim to reduce error and improve quality by standardizing processes.

These so-called process-management techniques took off in the U.S. in the 1980s at manufacturing companies desperate to compete with higher-quality Japanese products. But they have been widely applied across industries from retailing to financial services. Six Sigma, for example, was invented at Motorola Inc. in 1986 and popularized in the 1990s by former General Electric Co. Chief Executive Jack Welch, who preached its gospel across GE's businesses.

The systems can help boost quality and efficiency and help companies cut costs. A diaspora of former GE executives -- including W. James McNerney Jr., the former 3M Co. CEO who is now at Boeing Co., and Bob Nardelli at Home Depot Inc. -- helped spread the word. "Process management was sort of a universally held truth -- it was perceived to be universally good," says Mary J. Benner, a management professor at the University of Pennsylvania's Wharton School.

Now, some academics and consultants are raising questions about the limits of these systems. In essence, they say managers have stretched the techniques, by applying them too broadly to more creative areas such as research and new-product development. And some companies are rethinking the way they use the systems.

The critics say process management helps improve existing products and routines, but can hinder a company's ability to innovate. "For stuff you're already good at, you get better and better," says Michael Tushman, a management professor at Harvard Business School. "But it can actually get in the way of things that are more exploratory."

Ms. Benner and Mr. Tushman teamed up to study the subject. They found that quality-improvement or process-management programs can hamper a company's ability to respond to technological change, by effectively forcing employees to focus on honing routine tasks associated with the older technology.

In their study, the professors looked at the number and types of patents granted to photography companies from 1980 to 1999, a time of sweeping technological change as the industry began shifting to digital images from film. As companies used process-management techniques more, they generated more patents based on traditional technology, but fewer patents "based entirely on knowledge new to the firm," the study says.

Some executives agree that the techniques have limits in the business world. Christine Landon, director of executive and next-generation leadership development at Agilent Technologies Inc., says quality-improvement skills are important, but managers shouldn't rely too heavily on Six Sigma for tasks that aren't easily measured. As a maker of scientific instruments and testing devices, Agilent, of Palo Alto, Calif., is all about measuring the tiniest changes in the real world. But, says Ms. Landon, "when you try to apply Six Sigma too broadly to processes that don't have data, that aren't data-driven, that don't lend themselves to very specific metrics, that's really a stretch."

So Ms. Landon doesn't use Six Sigma to analyze her leadership-development programs, partly because the skills involved are "hard to measure," she notes. Plus, with Six Sigma, "you can go overboard. It's not an inexpensive thing."

GE, for example, estimates that it has spent hundreds of millions of dollars on Six Sigma since 1995, and says the program has helped it eliminate billions in costs. Nonetheless, GE has trimmed its use of Six Sigma in the past two years, concentrating it on functions that "touch customers," such as sales and marketing, but eliminating it for back-office operations such as payroll, says Gary Reiner, senior vice president and chief information officer.

GE executives don't believe Six Sigma hinders innovation, Mr. Reiner says. Instead, Mr. Reiner says, "we have a finite amount of resources and we want to focus in on those processes that touch customers." He notes, "there are times when Six Sigma is overkill."

As it turns out, Six Sigma may not even really be Six Sigma. The term is derived from a statistical formula for determining how "abnormal" a result is. Motorola translated this into a defect rate, as a way to measure quality. As published, Motorola equated Six Sigma-quality to no more than 3.4 defects in a million products, or a million times of repeating a process.

But in response to concerns raised by an independent business consultant, Motorola says this defect rate is actually closer to "four and a half sigma." True Six Sigma quality would be 1,000 times as tough -- no more than 1.2 defects in a billion products or repetitions. A Motorola official says it adopted the somewhat easier target because machines wear down over time, introducing more defects. So a process that starts out at Six Sigma quality will eventually drift away from that level, he says.

Write to Erin White at

This article taken in its entirety from the Wall Street Journal Website.  It can be located at:,,SB112709098674044484,00.html