January 2001

Vol. 3, No. 1

Six sigma and more: 
David R. Schwinn looks at the numbers

by David R. Schwinn

Last month, a note from Jay Bronec reminded me of a sign in Warren Tyner's office in the mid-1970s. Tyner was a Ford Motor Company chief engineer. The sign said, "In God we trust; all others bring data." 

That sign set the tone for all the design efforts of those reporting to him. That phrase also, certainly, sets the tone for Six Sigma efforts. Here, the goal is obviously to achieve six sigma capability. But the key question is six sigma capability of what. Which metrics should improvement efforts focus on? Let's review the wisdom of a few people who have been in this business of improvement for many years.

"The most important figures needed for management of any organization are unknown and unknowable." This powerful warning to anyone who manages by the numbers comes from Lloyd S. Nelson, Ph.D., director of statistical methods for the Nashua Corporation. W. Edwards Deming, Ph.D., frequently used that phrase to alert managers who might misunderstand his own message (W. Edwards Deming, Out of the Crisis, Cambridge, MA: MIT Center for Advanced Engineering Study, 1982).

Another teacher of mine, Jamshid Gharajedaghi, puts it another way. He declares that the important properties, such as love and success, are emergent (Jamshid Gharajedaghi, Systems Thinking, Boston: Butterworth-Heinemann, 1999). They cannot be measured directly. We must first define what we mean by success, for example, then develop a robust array of measures that serve as surrogates for the property we are trying to influence. This difficulty of figuring out what to measure is compounded by the old adage, "What we count, counts."

So what we choose to measure will influence what we do, and how we operate. But it will also influence what we don't measure, which can affect what we do and how we operate in an equally forceful manner. Jim Bakker, then vice president of manufacturing staff at Ford Motor Company, used to complain that the company's manufacturing operations did not perform preventative maintenance. When business was good, there was no time. When business was bad, there was no money. There was never a time that preventive maintenance was a key metric for plant managers. On-time shipment and labor costs were.

Myron Tribus further complicates this dilemma with his Perversity Principle:
"If you try to improve the performance of a system of people, machines, and procedures by setting numerical goals for the improvement of individual parts of the system, the system will defeat your efforts and you will pay a price where you least expect it." (Myron Tribus, Ph.D., P.E., "Quality First," Washington, D.C.: National Society of Professional Engineers (#1459), 1992).

The Perversity Principle reminds me of a phenomenon I observed early in my career at a major home appliance manufacturing plant. The hourly workers were frequently paid to go home early. You see, the materials manager was being measured based on the inventory level of purchased components, the quality manager was being measured based on outgoing quality, and the plant manager was measured on production efficiency. No one was watching the cost of paying the worker to go home early. So, when defective purchased parts were found and nothing was available to replace them because of low inventories, quality shut down the assembly line. When the line went down, the plant manager sent the workers home rather than sacrifice production efficiency. Deming calls this suboptimization. It happens whenever the Perversity Principle is ignored.

What should we do? Here are a few thoughts:

  1. Start with emergent properties such as success. Work with top management to operationally define what they want the Six Sigma effort to achieve. Don't let the existing metrics unduly influence this step. Remember, many metrics are used because they are easy to gather, not because they are important.

  2. Develop a set of metrics that can serve as surrogates. These may be metrics such as sales, inventory level, cost, defective parts per million, scrap, or cycle times. Make the array of metrics broad enough to truly represent what you want the Six Sigma effort to do, without being so broad that you lose focus. Make sure you ask what it is that you are not measuring that may result in unintended negative consequences. Make sure you're not creating measures and goals that, if achieved, suboptimize the overarching intent of the Six Sigma effort.

Deciding on metrics is serious business. Treat it that way.

As always, stay in touch. I'm at support@pqsystems.com.




Copyright 2000 PQ Systems.

Please direct questions or problems regarding this web site to the Webmaster.