A few months ago, I participated in an international conference and listened to a speaker describe her service organization's successful journey toward total quality management. At one point during the lecture, she showed a run chart on serious injuries. The average was about 15 per month. As I waited to hear how the organization had significantly reduced this high number, the next slide appeared showing a statistical control chart. The control limits had been plotted, and the speaker stated that this process was obviously in control. She then continued her lecture.
As I sat there dumbfounded, I couldn't help remembering W. Edwards Deming's famous line, "Putting out the fires in a hotel doesn't make the hotel any better!" After decades of quality management, how can we still be missing the point? Statistical quality control methods are wonderful tools for controlling quality -- when we have quality worth controlling. Of course we should put out fires in our hotels, but we should spend most of our energy finding out why we have fires and changing our performance to a level we wish to maintain.
In an article by a leading pundit of quality management in health care, I came across the phrase, "Quality directors in manufacturing companies have known for years that their most important task is reducing variation." Because the author is a close friend, I couldn't wait to visit him at his office so I could close his door and scream, "What do you mean our most important task is reducing variation? If we create a new drug that has 100-percent fatalities, is this good just because we now have zero variation?" Do we really want to minimize variation around a poor quality level? Or should we spend our time changing the level and then implementing the control system?
In the early 1980s, when companies began changing from inspection-oriented to more modern quality management methods, many organizations decided that massive work force training in statistical quality control was the correct first step. Unfortunately, the old saying, "if the only tool you have is a hammer, every problem looks like a nail," proved too true. People looked for applications of control charts rather than seriously considering what would add value to their organizations.
Another problem was that, even where appropriately applied, the feedback loop frequently was left open. Out-of-control points were plotted, but no corrective action was taken. Or people felt that the right action was simply to report the out-of-control point and hope someone else made the correction. The concept of self control was lost. There was a strong backlash against control charts and other quality management methods that were inappropriately applied.
Now we see that many organizations aren't using control charts even when they're the right tool. People have lost sight of what should be controlled and how. Internal -- and even some external -- assessors give good scores for the correct number of control charts, ignoring whether the charts are being used appropriately.
Developing a good quality control system is just not that hard. The first and most important step is to understand the hierarchy of control. What must senior management control? The strategic plan and goals, the capital budget, alliances and several other activities definitely are candidates.
At middle management and supervisory levels, department budgets, work schedules, cycle times, major projects and departmental work plans should be controlled, even though many of these recently have become candidates for self control or self-directing work teams. Most operational tasks should be controlled by the work force, and operating and administrative personnel. Whenever possible, the control should be automated.
The goal of quality control should be to identify process steps that best predict outcomes. By putting adequate controls upstream in the process, we eliminate the rework, scrap, reinspection and delays that occur when controls are added too late in the process.
A few years ago, I reviewed a quality team working through a large impact matrix. They carefully identified each process step that, if not properly controlled, had an impact downstream. They estimated the impact from each simulated lack of control and set priorities for what should be controlled. Then they reviewed the control plans, the sensors and measurements, and operator training at each key process. Where they found problems, they designed new control systems, bought new sensors and measurement devices, and provided good training to the operators. One by one, they eliminated the possibilities of out-of-control conditions causing harm later in the process. Few organizations have made that effort.
About the author
A. Blanton Godfrey is chairman and CEO of Juran Institute Inc. at 11 River Road, Wilton, CT 06897.
© 1998 Juran Institute. For permission to reprint, contact Godfrey at fax (203) 834-9891 or e-mail firstname.lastname@example.org.