What “Six Sigma” means is that a process should have output that is so exact, precise and consistent that it would have to be over three standard deviations out from what you’d expect before it was ever unacceptable to your customer. If your process is to make gun barrels, then applying Six Sigma to that process might mean that you are able to manufacture barrels which are so accurate that you’d have to do something really out of the ordinary–like pound your milling machine with a hammer during the fabrication process–before they were ever so much as a hair out of alignment.
The underlying logic is not necessarily to make the outcome better, but to control the process so that you get repeatable results. You don’t want to have to test every single gun barrel. You want such a perfect process that you almost never have to test.
Six Sigma revolves around the acronym DMAIC: Define, Measure, Analyze, Improve, Control.
Define: What are you trying to do? If you’re trying to improve survivability in car crashes, then stereo performance is irrelevant. You can’t fix everything at once, and there may be tradeoffs: more perfect products might increase cost, for example.
Measure: You can’t do anything without data, and the data has to be relevant to solving what you’ve defined in the first step.
Analyze: What does the data show? More importantly, is the data accurate? If somebody ever gives you a graph of data where the data points consistently go up, then down, then up, then down, they are lying to you. Real data is really hard to fake.
Improve: How do you make the process better? Maybe you move people’s workstations closer together. Maybe you get new tools. Maybe you do some training.
Control: Did your change work? And if it worked, how do you make sure it keeps working? You don’t test every circuit board you make. But you test enough to be sure that nothing else has changed with your process.
Is your process capable? That is, a very fine saw is capable of making very precise cuts, if calibrated properly. A chainsaw is not, and will never be capable of meeting a standard of high precision. Use the right tool for the job.
Is your process in control? Maybe when you grill a steak, you have no idea if it’s going to be overcooked or undercooked. From a Six Sigma perspective, it’s better to have a steak that’s always underdone: now you know for a fact that your grill just needs to be hotter, and that’s something you can improve and control.