Creating a Successful Six Sigma Program

A few days ago, I had the pleasure of renewing contact with a client that sent a few Green Belts to training several years ago.  That's all they spent on the program.  We coached all to successful completion, and since then, they've saved millions of dollars.

In this case, there was a key person in management who took responsibility for the program.  He was tenacious.  He kept new projects coming, and he required that people read the books, view the videos, and serve as project team members.  Knowledge was passed down from Green Belt to Green Belt.

The lesson is that you can't just pay money to a consultant and expect results.  Management has to take responsibility, and be fully engaged.  There is nothing that takes the place of management engagement.

How Long Does a Six Sigma Project Take?

The best time that I know of was 24 hours, with savings of $3.6 million.  That's a complete project, with all the usual steps.  To be fair, that was done by an experienced Master Black Belt, not someone doing their first project.

Difference Between Green Belt and Black Belt

We're sometimes asked, "What is the difference between Green Belt and Black Belt projects?".

The answer boils down to the type of tools necessary to solve the problem at hand.

Most real-world problems can be solved with fairly simple tools.  If you can find the process's input variables, prioritize them, study the important ones, take appropriate action, and constructively use Pareto charts, Process Behavior Charts, and can create a Control Plan, you can do a successful and meaningful Green Belt project.

If the problem requires advanced tools like the Student's T Test, the Chi Square Test, or Full or Fractional Factorial Experiments, it's probably a Black Belt project.

Health Care Green Belt Project

We've just wound up Advanced Green Belt training for a group in the health care industry.  Already, we have a project ready for certification.

The facility was taking much too long to complete incident reports, and Whitney reduced the median time by a bit more than 10X.

Good job!

Difference Between a Project Program and a Training Program

Don't get me wrong:  Good training in Six Sigma is a good thing.  But if you structure a corporate program as a training program rather than a projects program, you leave a lot of money on the table.

A Six Sigma program structured around projects measures itself by the number and quality of projects completed.  That is the thing that it reports to management.

A Six Sigma program structured around training measures itself by the number and type of trained "belts" it produces.

You can show the effects of projects on your Income Statement and Balance Sheet.  It's much harder to show the effects of just training.

Normality is Overrated

People are often overly concerned about having normally distributed data.  There are a few cases where having normally distributed data is important.  There are many more cases where distribution does not matter very much.  It's usually much more important that the data are drawn from a stable and predictable (same as homogeneous, or in control) process, but nobody seems to pay much attention to that.

Choosing the Normal Distribution to represent your data is an assumption.  Your choice of assumptions does not add information to the data.

If your data are stable and predictable, you can pretty well depend on the following rules holding true, regardless of how the data are distributed:

60-75% of the data will be within plus and minus one standard deviation of the mean.

90-98% of the data will be within plus and minus two standard deviations.

99-100% of the data will be within plus and minus three standard deviations.

In view of this, the conventional 68%, 95%, and 99.7% numbers for the Normal Distribution are a modest refinement of the general case.

Dr. Don Wheeler has another excellent article posted at Quality Digest, explaining how to use I-MR (he calls them XmR) charts to evaluate the precision and bias of a measurement system.  You can read his treatise here.

There are several tools for evaluating measurement systems, including Gage R&R and Intraclass Correlation Coefficient.  None are quite as simple and elegant as the one Don discusses.

Here is Don, quoting Churchill Eisenhart at Nation Bureau of Standards, back when that is what the organization was called:

Until a measurement process has been "debugged" to the extent that it has attained state of statistical quality control, it cannot be regarded, in any logical sense, as measuring anything at all.