Performance Improvement
H. James Harrington

Since the early 1920s, we've assessed process goodness by the degree that variation was minimized compared to the specification

Figure 1: Old Concept of Process Goodness in Service.

(see Figure 1). Control charts, process capability studies and six sigma have all been directed at minimizing variation. In the 1970s, IBM's magnetic-head manufacturing process required each process step to demonstrate a Cpk of 1.4 before a process's output was considered ready to be shipped to customers. We've been led to believe that the bigger the Cpk the better. The present popularity of the old Cpk approach, now named "six sigma," is certainly an outgrowth of this thinking. The repackaging of the 1950s process capability study under the name "six sigma" focuses on the importance of minimizing variation, not only in manufacturing but in all business processes.

 The concept of minimizing variation in manufacturing has been very well understood since Walter A. Shewhart's work in the 1920s. The key phrase in the previous statement is "manufacturing processes." I question whether tight frequency distributions are good indicators of process quality in the service environment. In today's service economy, we design our processes so that the output can be customized to each individual's requirements. In these cases, process quality is based upon the ability of its output to vary over the full range of the specifications (see Figure 2).

Figure 2: New Concept of Process Goodness in Service.

 For example, take the process of registering at a hotel. The specification might be: "Guest should be registered within five minutes, plus or minus four minutes." If the process had a sigma of one minute and an average of five minutes, it would be performing extremely well according to the old quality concept. However, if a hotel is interested in customer satisfaction, processing everyone in five minutes is probably not the right answer. If there's no one else waiting to register, the hotel clerk should process the tired business traveler who signed in at 11:30 in one minute and the young couple, visiting the city for the first time and seeking a city map and good restaurant suggestions, in eight minutes. It may even be good business to exceed the specified limits and talk with the couple for more than nine minutes if they wanted to continue the conversation.

 In the service economy, processes that can provide a wide range of variation in their outputs are preferable as long as the variation can be controlled to meet the unique needs of the specific consumer and the organization. In the service environment, it's desirable to have a wide distribution as long as the variation can be controlled and the service point is predictable. Of course, uncontrolled variation, even in the service industry, is harmful. The three primary measurements of a good service process are efficiency, effectiveness and adaptability. To be efficient, the process should consume a minimum amount of resources (less is best here). To be effective, the process should be able to provide output that exceeds the consumer's requirements (hitting the bull's eye is best here). To be adaptable, the process must be able to vary the way it performs over as wide a range of conditions as possible without adversely affecting efficiency and effectiveness (predictable variation is best here).

 In service-related processes, adaptability is the key measurement for outstanding performance in the 21st century. It's a new century, and we need to change some of our old beliefs. Let's make one of our new century resolutions to question everything that we've accepted as fact and create a new beginning.

 Here's an excerpt from H. James Harrington's Total Improvement Management (McGraw-Hill Inc., 1995):

 All organizations provide service to their customers, even such "hard goods" organizations as U.S. Steel, General Motors, Ford and Boeing. These organizations are not in the business to produce products. They are in the business to deliver a service to their customer. Ford does not exist to make Mustangs. They exist to service their customers by providing them with a means to move from one place to another. Every product manufacturer has at least some pure service processes that do not deliver tangible products to their customers, probably more than each of them realize.

 In their ninth new collegiate dictionary, Webster defines service as "contribution to the welfare of others" and "useful labor that does not provide a tangible commodity." Using those two definitions, one can see that the service industry is a major participant in the everyday business of the United States and the world.

About the author

 H. James Harrington is a former principal at Ernst & Young, where he also served as international quality adviser. E-mail him at jharrington@qualitydigest.com , or visit his Web site at www.hjharrington.com .