"Six-sigma" anywhere in the title of a book, article or website is guaranteed to draw attention. Businesses are urgently asking "What's our sigma?", knowing only that "6" is supposedly "good", and unsure of what is "bad". If you don't know a +6 from a -12 "sigma" then read this now, and don't make a "six-sigma" mistake.

RISK

When most people talk about 4, 5 or 6 "sigmas", they are often talking casually about a statistical calculation known as a "Z" score. For decision-making purposes, this is typically the next to the last step in calculating something that is actually useful!

Several years ago, a client asked me to predict the future! He said, "I have two welding work stations. They are each outfitted to perform the same kind of work, but each has been known to occasionally produce an improper weld. What are the odds that welds will be performed properly? What is the projected batting average of each machine?"

Pause, and realize that this is not the same as simply tallying up the historical performance. If you have never received a traffic ticket, does that mean we forecast "0%" chance of tickets in the future? Certainly not! An estimate of risk would have to take into account many things, including how and where you drive.

A calculation of risk (an abstract concept which you cannot "see") is a chain of reasoning with critical assumptions that must be checked at each link in the chain. Failure to do so produces results that are worse than wrong, they are arbitrary! ("Wrong" answers can be fixed.) In this chain of calculation, the prelude to the final answer is the "Z-score" ("Sigma") calculation. From this answer the final answer, risk, can be determined.

After comprehensive research, including a visit to the welding operation, this is the kind of report that I made:

I expect that Welder#1 will fail almost 32% of the time, whereas Welder#2 will fail only 4.5% of the time. To save money immediately, take the following actions based on this information:

1. Don't ever use Welder#1 unless you have so much demand, you have no choice.

2. If your engineers have to choose which station to replace or refurbish, spend the money on #1 to add capacity or on #2 to raise your capabilities.

3. If you need 1000 units welded, schedule 1048 if you use machine #2 and schedule 1465 if you use machine #1.

4. Make sure that your calculations of costs and profitability take into account the 48 (or 465) units of scrap and re-work; and modify your capital investment, product line, sales & pricing decisions accordingly.

I DON'T WANT MY "SIGMA", I WANT TO KNOW RISK!

Knowledge of the "odds" followed up with immediate action can make a business more profitable immediately without capital spending. We negotiate better contracts, make smarter scheduling decisions, and so on. A "sigma" does not help us do this unless 1) it is calculated in a credible fashion, and 2) we complete the last step in calculation by determining risk.

Have you had anyone offer to figure your "sigma" from your scrap report?
"Last month you had 5% scrap, so your "sigma" is approximately 2", they say.

You might ask, "And...what good is knowing this?" Well, someone might try to estimate your risk of scrap from the "sigma" of 2, but the estimate would be wrong!. The math would lead you to (No Surprise!) 5%!

Their mysterious "Sigma" calculation merely equates your recent scrap rate to risk. (Remember traffic tickets?) To anyone who really understands risk as a consideration in decision-making, this is an appalling equivocation.

LEVEL OF RISK IS STRATEGIC

Establishing an acceptable level of risk for today, and then choosing the means by which you achieve that level are strategic decisions. There is no single right answer for all companies.

Many business are quite profitable with a scrap rate of 5 per hundred, others would be broke at a rate of 5 per ten thousand.

You can lower your risk of scrap by primarily tactical (operational) means, as my welding example illustrates. You may also lower risk of defective product by spending capital dollars on technology. However, if your operation could not run the old technology correctly and consistently, then they probably won't do better with the new.

WHAT IS "SIX SIGMA"?

Although everything that allegedly has a bearing on business improvement, quality, productivity and clean living has tried to take a position under the "Six Sigma" tent, it is in essence a particular strategic method for managing risk.

Think about our welding machine again. A legitimate "Z-Score" of 2 implies a failure rate of just under 5%, and that implies 95% success. Suppose that our machine made a successful weld 95% of the time. If one weld was required to complete a product, then 95% of the yield from the welding process could be shipped to customers.

Suppose that three welds are required. If a single bad weld makes a bad product, then only 95% x 95% x 95% = 86% of the yield from welding can be shipped! If 10 welds are required, the yield is less than 60%!

Motorola is usually credited with originating the "Six-Sigma" strategy in electronics manufacturing where hundreds and thousands of sequential successes are required to make a successful product. Their goal was individual "in-process" manufacturing success rates of over 99.99%. That would result in a high yield of finished product, even though success depended upon all components working flawlessly.

Motorola calculated that the "Z-Scores" of these in-line processes would have to be about six with risk levels in the low per millions, if complex products were to work in spite of the enormous points of potential failure.

Motorola made a strategic business decision to intentionally over-engineer key processes and they profited from it. But remember, the goal is to identify and manage risk. There are many strategic and tactical options to that end.

AVOIDING "SIX-SIGMA" Mistakes

Before you jump on the "Six-Sigma" bandwagon, consider the following.

First, remember that original "Six-Sigma", in essence, was a strategic directive to engineering and not to operations. It establishes an extremely low level of risk as an engineering goal in the development of manufacturing processes. Over-engineering is not the only means of risk reduction. Your maximum benefit for the investment could be found in the improved operation and management of those processes.

Second, "Six" (and the level if risk that it implies) is not a magic number. It may not be right for you. Are you engaged in Rocket Science or Hog Farming? How are you measuring scrap or defect? Do you count the billion "ticks" each second of a computer clock as a billion successes if it works, or as a single failure if it does not?

Third, Get expert help if you want to baseline your present levels of risk. There are countless ways to err in the estimation of risk. With all of the hoopla regarding "Six Sigma", there is tremendous pressure on in-house Quality departments to produce a number that will be "acceptable", even if the means by which that number is concocted are worse than wrong.

Fourth, Remember that the goal is to manage risk and improve profitability. There are alternatives to an engineering/capital spending binge. Remember, "If all you know is a hammer, the whole world looks like a nail." Learn about alternatives that have roles for product engineering, process engineering, quality, operations and management.