Real-life quality problems are conundrums. Dorian Shainin realized that recognizing the distinctive characteristics of a problem was critical. He also knew that applying the right tactics was the key to the solution; however, many of the analysis tools of his time were not effective.

Further complicating problem-solving, not unlike today, was the lack of structure. The conditions in which challenging problems must be evaluated and solved is often unclear from the start. Added to that, problem-solvers generally work under stressful conditions.

The chief reason many problems remained unsolved was an ineffective approach. The approach was generally based on a set of tools that were less effective at unraveling tough chronic problems. 

Most generally, the traditional approach to solve chronic problems is the same as with common problems. It usually starts with a group of ‘experts’ brainstorming to identify the most plausible causes. From there, the focus is directed to the most likely cause(s). Many simpler problems can be solved with this approach. For the chronic problems, however, this approach often doesn’t work. The reason the actual cause did not surface was because...it was something yet unknown!

Shainin’s approach started with contrasting what worked and what didn’t. In essence, it starts with evaluating the behavior of the effect (the Y’s) first, to ask “what’s the difference between what works and what doesn’t?”   

In most situations, the majority of products perform as the manufacturer intends and, hopefully, only a small number fail to meet expectations. For this reason, the application of the Shainin technique of finding “what’s different” can be very effective. One key to solving product variation problems is to uncover the “Red X” by “talking to the parts.”

Shainin’s approach is a rigorous discipline for performance improvement in both engineering as well as manufacturing. The origins of this discipline go back to the 1940s when Shainin and Dr. Joseph M. Juran recognized that the Pareto principle, initially attributed to the distribution of European wealth, also applied to the causes of variation.

Shainin ultimately concluded that problems were not caused by many X’s but by one dominant X. If the Red X could be isolated, the problem could be solved. The approach was to make decisions based on observable facts, not opinions as often is the outcome of brainstorming. Shainin’s method was to “talk to the parts” by witnessing and evaluating what is really happening but that is easier said than done.

To illustrate this point, I was involved with shifting noise in a certain transmission model. Normally very silent, an occasional unit would exhibit noticeable (to the ear) noise. The noise was also detectable during end-of-line testing. The baseline of the problem could be plotted on a graph (or tracing), revealing the variation in the noise level over a number of manufactured units.

The human ear is not necessarily a reliable system so to determine the actual noise level was being generated by the manufacturing process, a gage reliability and repeatability (GR&R) determined the measurement system was not contributing to the problem. Therefore, the variation in the tracings did reflect good (silent) and bad (noisy) product.

Using traditional methods, noisy units were disassembled. All components were carefully measured but engineering specifications were all within tolerances. Units were carefully reassembled and retested, but the noisy units remained noisy.

Perplexed, the next step was to select samples of product on the extremes; therefore, a few very quiet and a few very noisy units (determined by tracings) were selected for evaluation. The decision was made to use what Shainin referred to as component search. Units were disassembled, carefully re-assembled and re-tested. The result was no change to noise levels; therefore, the noise was unrelated to the assembly process and related to component parts.

In the next phase, to identify the offending parts, the units were disassembled and parts ‘swapped’ from the bad units to the good. After reassembly, the units were again tested with noise levels measured accurately. This experimental process revealed that a certain component could switch a good (silent) unit to a bad (noisy) unit and vice versa. It could also be verified by ‘swapping’ back the same components. 

After a series of paired comparisons, the elusive problem was isolated. The problem was a shaft which had extreme levels of variation in the supplier’s manufacturing process. The culprit wasn’t easily isolated, but once discovered, there was a relatively simple solution.

If you’re confronted with a difficult challenge, I’d encourage the application of Shainin’s System (SS). You just might find success.