What is Six Sigma Quality, and why is it a hot management topic?
Six Sigma Quality is a movement that inherits directly from TQM, or Total Quality Management. It uses much the same toolset and the same concepts. Six Sigma Quality has two new emphases which are its distinguishing characteristics:
1) Six Sigma Black Belts - well-trained experts in quality, process improvement, and statistical process control - who work within companies as "problem-solvers for hire". They lead process improvement projects, and focus on areas which will have the highest impact on the bottom line.
2) A focus on reducing variation to very low levels.
Jack Welch, the energetic chairman of GE, has been Six Sigma's most influential advocate. Other companies, notably Motorola and Allied Signal, have been incubators and proponents of the movement. Mikel Harry is its most colorful champion. The consulting and training firm he founded, Six Sigma Academy, has, become the most well-known educator of Black Belts. Many other traditional quality consultancies have been quick to follow suit, including Six Sigma Qualtec, the Juran Institute, and Oriel.
The name - Six Sigma - wants some explaining. Imaging that you are weighing bags of potatoes as they come out of a bagging process. The bags are supposed to weigh 10 pounds, but the actual weights will vary. If they are overweight, you are giving away potatoes. If they are underweight, you are ripping people off. So...you record the weights, and use some software to construct a histogram of the distribution. You would hope that the distribution would be centered on 10 lbs., and that there wouldn't be long tails on either side. If your specification calls for all bags to exceed 9.5 lbs., and to be less than 10.5 lbs., you can draw these spec limits on the histogram.
So... you're measuring along, and plotting the histogram, and you come across a bag that weighs 9.4 lbs. It is out of spec. It may cause trouble with customers if you ship it. What do you do? How many out of spec bags do you expect to find if you measure 1000 bags? You need some way of predicting this. That's where statistics come in. You can find the average, or mean weight of a bag of potatoes from all the ones you've weighed. You can calculate a standard deviation, too, which gives you an idea of how much variation there is around the mean. If the standard deviation is high, that means you have a lot of variation in the process. Here's where the sigma comes in...the Greek letter sigma is usually used to symbolize the standard deviation in statistical equations.
With enough data, you can try to fit a curve to your data - drawing the line that best approximates the mathematical function that really describes what is going on in the process. The art of curve-fitting is an arcane one, and not one we need to go into here. Let's just take a normal curve as an example of one that we might decide to use, if we went through a curve-fitting exercise. If our histogram can be well-enough described by a normal distribution, then 68% of the bags we measure will weigh within one standard deviation, or one sigma, of the mean value. If the mean is 10 lbs., and the standard deviation is 0.2 lbs., then 68% of the bags would weigh between 9.8 lbs. and 10.2 lbs.
Again, if this is a normal distribution, we would find that 95.5% of the bags weighed within 2 sigmas, or 0.4 lbs. of the mean. If we look at the mean plus or minus 3 sigmas, or 10 +/- (3*0.2), we would find that 99.7% of all bags would weigh between 9.4 lbs, and 10.6 lbs. A very few, just 0.3% of all bags, would weigh less than 9.4 lbs or more than 10.6 lbs.
If this is the case, though, our process is wider than our specification limits. Some of the bags weigh less than 9.5 lbs, and some weigh more than 10.5 lbs. Does that matter? It depends. With potato bags, maybe not. But what if we were measuring something with fine tolerances, or something expensive, or something where precise mixtures were critical. We wouldn't want to be finding instances where our process was not producing outputs that not in spec limits.
The Six Sigma Quality movement takes this very much to heart. In fact, six sigma advocates believe that for many processes, there should be six sigmas between the mean and the specification limits, so that the process is only making a few bad "parts" in every million. You can, of course, do that by relaxing the specifications, but that isn't usually the way to please customers. Instead, the variation in the process needs to be driven towards zero, so that the histogram gets narrower, and fits more comfortably inside the spec limits.
Clearly, to get an accurate view of your critical processes, you need to have people who understand variation and statistics. The Black Belt training spends a lot of time on this. Software tools, such as control charts and histograms, are harnessed. So, too, are the tools of quality improvement, teamwork, project management, and creative thinking. Root causes of variation are explored, and the classic Deming PDCA cyle is used to plan improvements, try them, check to see if they worked, and standardize on them if they did.
We don't see that a lot of the Six Sigma methodology is new. It combines elements of statistical quality control, breakthrough thinking, and management science -- all valuable, powerful disciplines. We are happy that in this new movement, the time-tested tools of quality and process improvement are getting renewed high-profile attention and achieving excellent results.