RE: Six Sigma & Deming Philosophies!

[Rip Stauffer <rstauffer@bluefirepartners.com>]

Hello, All,

OK, let me preface this with the following caveat: the firm I work for does
help organizations with Six Sigma deployment. If this introduces too much
bias into my remarks, then just dismiss them. I'm hoping that I can be
somewhat objective, here. I am a long-time student, advocate (and I hope
practicioner) of the Deming Philosophy, and did all my graduate research on
Shewhart, Deming, and Analytic/Enumerative studies. 

I was asked to investigate Six Sigma several years ago for a much larger
institution. After attending a seminar, I was much unenamored of the Six
Sigma approach, because of a number of things (this is not an exhaustive
list):

1. To me, it looked like Crosby's "Zero Defects" on statistical steroids
2. No one could explain the "1.5-sigma shift" to my satisfaction
3. No one talked much about quality; it was mostly about cost-cutting and
cutting cycle time (whether it needed to be cut or not)
4. It had no real philosophical underpinnings

By the time I was ready to retire from the Navy and move into the world of
civilian business, a couple of things were readily apparent:

1. No one seemed interested in hiring a "Deming Philosopher."
2. If you wanted to talk about a job in Quality, you needed to be able to
talk about Six Sigma.

I also talked to a couple of people in the DEN about Six Sigma. They told me
that I should learn more about Six Sigma, to see what its flaws were and see
whether it could be improved.

I was lucky enough (actually through the DEN) to get hooked up with Lou
Schultz, a friend of Dr. Deming's, who had a small consulting company that
had been spreading the Deming Philosophy and helping companies since about
1984. He asked me what I knew about Six Sigma. I told him about the
experience I had had above. Lou asked whether I was willing to learn. I
said, "Of course."

One thing that became immediately apparent as I began attending conferences,
talking to people from organizations immersed in Six Sigma and to other
consultants "doing" Six Sigma, is that "Six Sigma is not Six Sigma." It's a
little better defined than TQM, but not much. 

Many organizations do try to use it merely to cut costs. Some consulting
firms, which until recently were much more about accounting than quality,
are more than willing to perpetuate this. The worst aspect of Six Sigma, and
the one that may kill it, is the proliferation of "Black Belts" who never
had anything to do with quality until they were forced into it to meet some
mandated goal in their performance appraisal. They heard what they heard,
and now are out in the marketplace with "Black Belt" as the most marketable
bullet in their resumes. I have seen and heard some incredible nonsense from
some of these folks.

I would like to point out, however, that Six Sigma is not all bad.
(Attention: here's where my "bias" may begin to show.) 

The first point I would like to make is that Six Sigma, as a methodology for
studying cause systems to find improvements that get us on target and reduce
variation, has some merit. It does not apply everywhere, certainly, but it
does work very well when applied well, like anything else.

The second point is that it is "Philosophy-neutral." I have, I think, been
somewhat successful in helping some organizations begin to appreciate the
SoPK because of the way we approach Six Sigma. I use the Red Bead, the
Funnel, discussions of Systems and Psychology, discuss theory of knowledge,
in every class, and we work hard to make sure the students walk away
challenged by these concepts.

I struggled for a long time (and still do) with the 1.5-sigma shift. I have
some clients who are not interested in the "Six Sigma Metric." For them, we
talk about yield, and that works pretty well. I know a bit more now about
the origins of this "shift assumption:"
1. It came from Motorola, a company producing extremely high-volume products
and having trouble finding subgrouping schemes that would enable them to
confidently track performance in those processes.
2. Apparently, they took a lot of data and discovered that there were times
that the process would shift for a while before it was detected.
3. It seems that the research indicated that 1.5 sigma was a pretty good
number for an overall summary or estimate. They (and most of the rest of the
Six Sigma world) now gig a process 1.5 sigma in order to have the same scale
for comparison.

Now, a couple of points. I have a Ph.D. Statistician (also long-time Deming
advocate) friend who worked at Motorola. He shared with me a paper he wrote
that stated that had the folks at Motorola used all four Western Electric
Zone tests, they might have found many of the shifts earlier. When we
discussed this paper, though, he pointed out something that had not occurred
to me before: If you look at these tests, some of them do have a lag. 

Suppose you are taking four samples per day from some 10,000-unit-per-day
production line and averaging them. Suppose that you know (because you have
developed this practice over time, based on sound theory and lots of data)
that this will reveal special causes. This might be pretty good SPC
practice, given a particular process, and is the scheme that was used at
Tokai Rika in the process made famous by Wheeler's "The Japanese Control
Chart." Suppose you find a run of eight above the centerline, and discover
the cause, and fix it. That would mean you have gone eight days with the
process shifted somewhat, and have probably produced somewhere around 80,000
parts with the mean off-center. 

So does shift happen, even in well-controlled processes? I think it probably
does. At least it has in most of them I've seen. To me, this is not
necessarily a cause for alarm. It is the reason we have to constantly
monitor and look for special causes. It is the reason we have to keep
working to center processes on the nominal and reduce the variation. But the
process will shift sometimes, and run for a while before we notice it.

Now, using 1.5 as a universal swag seems excessive, and you can argue all
day long about whether we actually need some report card metric to compare
processes. I think that in the hands of the knowledgeable, that type of
metric might be useful to know what needs more attention in an organizaton
that can only devote resources to a handful of problems at a time. I know
that's a big caveat (hands of the knowledgeable). Most people who understand
statistical thinking don't put a lot of credence in the Six Sigma metric as
anything like a real number. You have to ask a lot of questions about that
number to know whether you can compare it with another one. You are using
one index to compare, say, a count of errors with the capability of some
continuous process. 

I do often wish that Motorola had not saddled us with this "metric." I try
to get my clients to adopt whatever is most meaningful for them for
comparison, and to track their progress. 

As far as Six Sigma being used only to cut cost...it can and has been used
that way. Organizations that stick with it eventually find out that that
approach can only take them so far, though. The good thing is that if they
have learned the power of reducing variation and optimizing their systems,
they find that the principles, tools and methods they have learned apply to
much more. 

One of the good things about money, though...executives seem to understand
dollars much better than they understand Cpk. My experience has been that
they actually do "give a damn about profit," but they don't understand the
chain reaction that creates it. If we can get their attention by improving
their financial situation up front, it's usually much easier to get them to
understand the importance and vastness of the unknown and unknowable
numbers. If we go in up front only talking about the unknown and unknown and
unknowable, we won't last long.

One other thing...is it a throwback to MBO? This is a troubling point for me
as well. Again, though, it depends on how it's used. Let me talk about two
scenarios:

Scenario one - Customers want on-time delivery of some component that my
industry supplies. To most of these customers, "on-time delivery" means in
the Kanban for that component within 20 minutes of requested time, and they
are frustrated that no one seems able to meet that. Right now, we are able
to deliver within a 60-minute window. The industry standard is 50 minutes. I
could ask an improvement team to beat the industry standard, and gain some
market share, I could ask them to meet the 20 minutes, or I could ask them
to hit a "stretch goal" of 10 minutes.

Scenario two - As a call center manager, I have been given a goal by my VP
of ops to reduce operating cost by 25% this year. I know that I pay a lot
for each of my call center reps, and that each minute they talk to customers
costs me something. I set up an improvement team to cut call center costs by
25%. I tell my boss about it, and he says, "Great...I'll take that money out
of your operating budget for next year."

I don't know...maybe there's no difference between these. I have no problem
with a stretch goal in the case of scenario one, if the manager understands
that it's a stretch, and is willing to give the team credit for whatever
they do. Certainly, the 85% solution works in that case. A stretch goal can
be good for an improvement team, because it often forces them to practice
creative thinking. They might get 5% by luck, just doing what they've always
done a bit better; but they wont get big improvements that way.

Unfortunately, I have seen executives with little understanding try to use
Six Sigma as a club, as in scenario two. In this case, it's just another
tool to use badly for the wrong purpose. If they didn't have Six Sigma, they
would use something else. 

A final point: I didn't really come to defend Six Sigma in this forum. I
don't want to get into an extended argument over any of these points. I did
want to explain some things that I have learned in the hope of providing
some amplifying information and maybe raising some new questions around some
of the issues that always come up when we discuss this. I have found Six
Sigma to be useful as one set of tools. What's emerging as Lean is another
set. Traditional SPC is certainly another powerful set. I believe Six Sigma
can be used well or badly, like any other set of tools. 

I hope this proves useful to somebody. BTW, there is a link on our website
to a pretty great presentation on Six Sigma and Quality Culture Change at
some Mayo Healthcare facilities:

http://www.bluefirepartners.com/performance

Just click on the link to "A Six Sigma Deployment Story" Dr. Rupp has some
excellent things to say about change challenges.

Best regards to all,

Rip 

Rip Stauffer
Senior Consultant
BlueFire Partners
150 South Fifth Street, Ste. 1300
Minneapolis, MN 55402
612-244-1027