Constraining Your Improvement Activities to Manufacturing Processes

In this installment of my arbitrary and capricious list of the Top Ten Stupid Six Sigma Tricks (SSST), let’s talk about an error that is perhaps less frequently made now than it has been, but is still common. I call this error SSST No.6, constraining your improvement activities to manufacturing processes.

In an oft-repeated quote, Bob Galvin (the former CEO of Motorola) said, “The lack of initial Six Sigma emphasis in the nonmanufacturing areas was a mistake that cost Motorola at least $5 billion over a four-year period.”

These are two areas in which this SSST appears: the nonmanufacturing areas within a manufacturing business and the nonmanufacturing sectors themselves.

Nonmanufacturing areas within a manufacturing business

People usually have the feeling that within a manufacturing business the biggest opportunities for saving or making money are in that part of the business that manufactures. I won’t deny that there remain many inefficiencies and improvements to be found there. Keep in mind, however, that in most businesses the manufacturing turnip has been squeezed for efficiency improvements for a long time. Further improvements are harder to find. So why do people still have the impression that there’s more easy money to be gained here?

I think it’s because we have a lot of practice measuring the costs of manufacturing processes. These costs are concrete; we can see the money go in and something (salable or scrap) come out. Accountants even group together all these operating costs as a line item—cost of goods sold (COGS)—to separate them from other costs such as general and administrative (G&A) costs.

As a consequence, if Black Belts make improvements that result in $1 million savings, it’s (theoretically) easy to see the cumulative effect, either on total costs or on a per-unit basis.

On the other hand, making improvements in the “other” line item is more nebulous, because changes in that area may not affect, and could even increase, the G&A costs. Savings here may be much harder to identify and allocate to improvement activities.

For example, let’s say a design-for-Six-Sigma Black Belt is working with procurement to build a supplier quality assurance system. In the early phases, using statistical sampling and maybe some experiments, the team shows a link between variability in a raw material characteristic and an ongoing surface defect on the product. Interestingly, she finds that the variability of the material supplied by FBN Inc., the current supplier (whose CEO is her CEO’s golfing buddy), increases the rate of a surface defect on the final product. Their competitor’s raw material, supplied by ACME, would actually decrease it. However, ACME is twice the cost of FBN and their CEO is too busy working on continuously improving quality to play golf.

So, she finds that a case can be made to reduce the total cost by purchasing ACME. She bravely makes that recommendation and, assuming she remains employed and they implement the finding, the COGS goes down significantly, even with the higher raw material cost, because she has eliminated a long-standing reason for scrap by changing the raw material vendor. However, an ongoing supplier quality assurance system needs a budget to maintain and that cost gets directly applied to G&A. Investors and the board of directors might look at the ratio of G&A to sales as a measure of management efficiency and conclude things were getting top heavy (see SSST #8for more discussion on the perils of overhead). This is the case even if the increase in G&A is small compared to the savings of that one vendor change, because these are two separate line items to an accountant or investor.

To avoid this, it’s critical for management to have horizontally integrated metrics to fully appreciate the effect of improvements in the nonmanufacturing part of a manufacturing business.

Six Sigma outside of the manufacturing sector

According to Douglas B. Cleveland of the U.S. International Trade Administration, as of January 1999 services made up about 80 percent of the U.S. economy, or about half of the country’s gross domestic product (GDP). Yet even now, very few service and nonmanufacturing companies have embraced even the most basic quality concepts that were, after all, originally developed in manufacturing.

In fact the potential benefit from using Six Sigma tools in the nonmanufacturing sector is huge. In my experience, there’s money to be made here and I consider nonmanufacturing to be the biggest future growth area for Six Sigma tools.

Why is that?

Those of us with manufacturing backgrounds see things in terms of processes, and we have a number of basic tools that we can bring to bear in examining and improving processes. We understand that the outputs of one process are inputs into another process. We have been trained to use these tools to improve processes and we know everything is a process.

On the other hand, people in nonmanufacturing haven’t typically been brainwashed this way. Many tend to see the outputs of the business as the result of a number of occurrences, including, in the words of W. Edwards Deming, “willing workers just doing their best," and people doing what needs to be done. In a sense, I believe that many people in these sectors consider the business “emergent” from their actions rather than a product of a coherent design. If one views a business this way, when a problem occurs, it’s baffling to even begin to think about how to fix it. If I didn’t do something wrong, then why should I be involved in the investigation or solution of the problem?

Don’t get me wrong. I’m not saying that nonmanufacturing workers are any less smart than those in manufacturing. I know some very savvy people in this sector, but without the paradigm of viewing a business as a number of interacting processes, the human brain just has little traction to find solutions.

This is one reason Six Sigma hasn’t penetrated the service and transactional businesses as much as it could—it’s seen as a manufacturing thing that doesn’t apply to them, rather than a process thing that does. (See the Six Sigma in service article at http://www.qualitydigest.com/may03/articles/01_article.shtml)

I see a bright side to this, however. Once people start thinking of their activities as a process, they can get huge improvements just by using some of the basic quality tools. One Black Belt from a very large transactional business was able to save $15,000 per month by examining application processing with a flow chart. In the near term for these companies, it’s not just that there’s low-hanging fruit, it’s that there is fruit lying on the ground just waiting for someone to trip over it.

Differences between manufacturing and nonmanufacturing

Another reason that Six Sigma hasn’t spread to more nonmanufacturing businesses or nonmanufacturing areas within a business is that while it is true that “a process is a process,” nonmanufacturing businesses do, by their nature, tend to generate a different level of data than in manufacturing. The sad truth is that most Black Belts aren’t trained in how to handle these data.

WARNING: tech discussion ahead! In the event of a water landing, your laptop can be used as a flotation device.

For example, those of you who are Green or Black Belts, raise your hand if you think that ordinal survey data can be analyzed with a t-test to determine differences in responses between two groups.

The answer, as I am sure you already knew—I was trying to catch that other person over there—is that it cannot. Calculating a mean or doing a t-test on ordinal data is completely inappropriate, and you can end up with the wrong answer if you do. However, most Black Belts I have met wouldn’t even ask what level the data were, much less know that they needed to do a Wilcoxon-Mann-Whitney test. In manufacturing, we frequently have interval or ratio level data. On the nonmanufacturing side, procedures for nominal and ordinal data are more frequently needed.

I notice more and more Six Sigma providers offering service- and transactional-themed training, and this might be a step in the right direction. However, as a rule, the training is shorter and less statistically intensive. This might be because of the mistaken impression that service and transactional workers are “less data proficient” or, less charitably, that they are “data-challenged.” Ironically, these are the people who will benefit the most from nonparametric analyses, which are frequently not even included in manufacturing Black Belt training. The true case is that, even more than manufacturing Black Belts, service and transactional Black Belts need these advanced tools and the depth of understanding to use them correctly so they can gain the higher, sweeter fruit after the basic tools have been applied.

Effect of SSST #6

So who cares? As we in the United States watch our economy move from mostly manufacturing to mostly service, all of us should care. At its best, Six Sigma takes real money that would have been completely lost and puts it back into the business, which in turn means the business has a higher efficiency in turning resources and effort into products and services. The service and transactional industries haven’t typically been viewed as processes, and as such, there are likely massive inefficiencies in them, probably similar to manufacturing in the 1970s, and there are very few people tasked and trained to capture that.

This is one stupid Six Sigma trick whose dam of resistance I hope has started to leak and a deluge may be inevitable. Manufacturing companies have begun to apply the tools across all parts of their companies and service companies are increasingly interested in finding out more about Six Sigma. As I have said before, the ones that get there first, win.

And listen, there’s an elephant in the room that everyone sees and tries to ignore to stay sane. Over the next 10 years, health care spending will approach $4 trillion or 20 percent of the GDP, according to the National Coalition on Health Care. That’s one service industry we all use, sooner or later. Imagine the effect on each of us if we could use these tools to improve that industry’s efficiency by a modest 20 percent. It would be massive.

But I could be wrong.