The truck breaks down. The router bearing blows. Lenny calls in sick the morning of the big installation. Bad luck? Gremlins? Or bad systems that can be fixed? How do we know?
Statisticians have something they call the "gamblers fallacy." I saw it in action at the Aladdin. I was watching the roulette table. The two men beside me could barely contain their excitement. Black had come up seven times in a row. They bet big on red. Black. They doubled their bet. Black. Again and again, huge piles of chips on red. After black hit for the 12th time in a row, they walked away, out of money, disgusted, and the Aladdin had $5000 that five minutes ago was theirs. I stayed and red finally hit after the fifteenth consecutive black number. "Wow," I said to the croupier.
"It happens," he said, his face a bored mask.
It happens because in a random system the odds for the outcome of the next occurrence are always the same — 50/50 (almost). The "gambler's fallacy" is the belief that, in a random system, previous patterns or history will have any effect whatsoever on the next outcome.
The point? Our businesses seem to be filled with random elements — those gremlins — that we can't get rid of. How do we know which are just part of living in an imperfect world — bad luck — and which are the result of errors and omissions in our business systems? This information is crucial, especially as we improve our businesses, system by system. If there really is a point at which we've hit the best it's going to get, then further work will be pouring money down the same rat hole as our two gamblers.
We need to know: Which can be fixed and which can't?
For instance, we may pride ourselves on our customer service. We have worked on it, improved on it, and have developed what we believe is a pretty darn robust system of keeping happy customers. Yet sure as rain, here comes the customer from hell whom nobody or nothing will ever satisfy. Have we overlooked something?
Total Quality Management provides us with a nifty tool to give us the answer, a tool that is as simple as its name is complicated: Statistical Process Control (SPC).
Here's how it works:
Identify a system.
How about "selecting our customer base?" Every business selects its customers. Nobody can be all things to all people. The best businesses do it consciously, with a strategic plan that is tied to marketing but gets input from the whole enterprise. But even a business without a plan self-selects a certain kind of customer based on what part of the price-service-quality triangle they emphasize. Even the owner's personality can be a powerful selection tool.
Define quality for that system.
Quality here might be "having a happy customer," but to make this useful, we have to break it down into smaller parts, and then nail down each part into something we can a) observe, and b) count. The list might look like this:
Chooses another supplier
Define a period of time.
For example, disputes per (day, week, month).
Make a graph.
One axis will be the number of occurrences, and the other will be the units of time.
SPC graphs always start with a high number. This is because they are tools to make things better. If you are not having a problem in an area, why go to the trouble of making a graph on it? In our example, this owner is spending a LOT of time in shouting matches. Morale and even cash flow are suffering. He and his staff, however, are on the right track — the graph shows it. They are brainstorming, implementing system changes, and the results are there to see.