One Size Does Not Fit All

Identifying the right improvement methodology

by Roger W. Hoerl and Ronald D. Snee

The need to improve is ever present in all endeavors and will continue to be so. We live in a dynamic world. As predicted in the second law of thermodynamic and entropy, the world will continue to change. This change produces a need for improvement; we have learned that if left alone, things get worse—not better.

Lean Six Sigma is generally considered to be the best available approach to improvement: true for a lot of problems, but not all problems. In improvement, one size does not fit all. We have argued a holistic improvement approach is needed, with the four critical building blocks being: leadership, top talent, infrastructure and holistic improvement method,1-3, meaning it can handle any problem.

The problem, of course, should define the improvement method. To help get to the correct characterization of the problem, it is helpful to evaluate two critical variables: whether the solution to the problem is known and the complexity of the problem.

When presented with a need to improve, the first question we must ask ourselves is whether we know the solution to the problem. Examples of a solution-known problem might be to pave the parking lot or install a new computer system. We assign a project manager to implement the solution.

Knowing the problem solution does not mean the solution is easy or unimportant. For example, a person might know why he or she is overweight, and what must be done about it. Keeping weight off is not easy, however. A key question is, “How?” That is, “How to implement solution?”

If the solution is unknown, we don’t know why we have this problem. The key question is, “Why?” That is, “Why is this problem occurring, or what is the root cause?” Here, a problem-solving approach and mindset is needed.

Both kinds of problems are important to the success of an organization’s improvement initiatives. In fact, a blend of solution-known and unknown problems is the rule rather than the exception. The principal difference is that approaches to discover the best solution are needed in solution-unknown problems.

Problem complexity

Another critical consideration is the complexity of the problem. Problems with low complexity are typically isolated problems: problems involving a special cause. The important question here is, “What went wrong?” You find the root cause, fix it and return to normal.

Problems with high complexity typically involve the entire system and involve working on common-cause variation: There’s no special cause to fix. Here, an improvement approach such as lean Six Sigma may be appropriate.

Taken together, these two variables create a matrix (see Figure 1) that is helpful in identifying an appropriate problem-solving method.

Figure 1

The problem is categorized into one of four quadrants in the matrix that helps identify the best solution method. Note that the listed methods are illustrative, but not exhaustive; other methods could be added. Let’s look at some examples that illustrate how the matrix is used.

Case 1. Last year, we presented a case study in which a company that manufactures flushable baby wipes found that its product was clogging septic systems.4 Obviously, this was a serious problem for the company, but was also clearly a special cause: This hadn’t happened in the past and occurred suddenly. This suggested that the problem had relatively low complexity. However, the solution and the root cause were unknown.

This placed the problem in quadrant two. The company chose a team problem-solving approach, using techniques such as brainstorming, Pareto charts and run charts. The root causes and appropriate solutions were found using these basic tools, resolving the problem and returning the process to normal.

Case 2. One of this column’s authors was once asked by an IT professional to help him on his Six Sigma project. It turned out that he wanted to install an Oracle relational database to replace a manual system. The author asked him whether he had done this before, and he replied yes. So the author asked the IT professional whether he felt he knew how to do this, and again he said yes.

Looking at the matrix, this was clearly a solution-known problem. Even though he had done it before, the problem had high complexity because it could easily be mismanaged with serious consequences. Also, the details of how to deploy the database and phase out the manual system were not obvious. It was clearly process improvement versus problem solving; the intent was to take the process to a new, higher level of performance. Therefore, it was a quadrant three issue. A reengineering approach was used, including lots of mistake proofing and project management rigor.

Case 3. The same author also was involved in the design and building of a new, state-of-the-art paper mill, which involved the application of recent developments in manufacturing excellence. This was a solution-known problem because the team knew what it wanted to build, but figuring out how to build the facility and develop detailed work processes were quite complex. Therefore, we view this as a quadrant three project.

The team used an approach similar to what today would be called lean manufacturing. The mill was designed and built using these lean manufacturing principles and methods.

Case 4. In our book, Six Sigma Beyond the Factory Floor, we discuss a project that involved development of a model to predict corporate loan defaults in the context of statistical engineering.5 There was no special cause to fix, and clearly predicting companies that are likely to default is highly complex. Further, there was no known solution, which put this project into quadrant four.

The team used the Six Sigma approach—specifically, design for Six Sigma—to attack the problem. Further, there were some subprojects to the overall project. For example, one team was responsible for identifying appropriate data sources as the basis for predictive models. This task was solution-known, low complexity (quadrant one).

It was still important, and nontrivial, but the team knew what was needed (good data covering defaults and nondefaults for more than 20 years), so the solution was known. There was a problem to be fixed: lack of data. The team knew the sources to visit to obtain it. They used a Nike approach: Just do it. Expertise was needed, but a formal approach wasn't. Another team had to develop a formal definition of “default” because no generally accepted definition existed; they used a WorkOut to do this (quadrant one again), pulling a team together to discuss this problem and work out a solution without any formal tools.

Case 5. Author Thomas McGurk wrote about a major corporate change initiative to improve the supply chain capability of a large biopharmaceutical company.6 In looking at the details of this initiative, we believe the overall effort included elements of all four quadrants from Figure 1. For example (see Figure 2):

Figure 2

  • Conducting bottlenecking analyses of equipment and people (quadrant two).
  • Improving process yield and reducing batch release time (quadrant four).
  • Creating new quality specifications for critical raw material (quadrant two).
  • Creating a 50-parameter process monitoring measurement system to aid the control of the manufacturing process (quadrant three).
  • Leadership training for the management team and operating skills training for the process operators (quadrant one).

It is not surprising that all four quadrants are used because this problem was a major initiative conducted during a one to two-year period. Capacity was increased more than 50% for two major products, ensuring the blockbuster drug supply would satisfy market demand.

These applications of the matrix and other experiences show that:

  • Many, perhaps most, problems will fall into one or more, but not all, of these quadrants. This is natural and to be expected.
  • After we know the quadrant or quadrants the problem occupies, the possibilities for identifying the right improvement method are greatly enhanced.
  • The problem-solving speed and accuracy is increased.
  • Probability of success is increased.

Many practitioners have told us over the years that they understand different methods, such as lean Six Sigma, kaizen events, reengineering and WorkOut, but they are confused as to which improvement approach is the best. They also wonder if one method is best for all problems.

As noted previously, we feel that the approach should be based on the problem, not the other way around, and no one method is universally best. We have found this matrix works well in guiding practitioners in diagnosing the best approach to a given problem, and we recommend it for your consideration.

Of course, it is not a prescriptive system. It should serve as a guide and not dictate the approach to be taken. Further, it is not exhaustive; many more methods could be added to the matrix. Use of such tools to guide improvement efforts is an example of statistical engineering, a topic we have discussed in detail in earlier Statistics Roundtable columns.7-9


  1. Ronald D. Snee and Roger W. Hoerl, Six Sigma Beyond the Factory Floor, FT Prentice-Hall, 2005.
  2. Ronald D. Snee and Roger W. Hoerl, “Integrating Lean and Six Sigma: A Holistic Approach,” Six Sigma Forum Magazine, May 2007, pp. 15-21.
  3. Ronald D. Snee, “Digging the Holistic Approach—Rethink Business Improvement to Improve the Bottom Line,” Quality Progress, October 2009, pp. 52-54.
  4. Roger W. Hoerl and Ronald D. Snee, Statistical Thinking; Improving Business Performance, second edition, John Wiley and Sons, 2012.
  5. Snee, Six Sigma Beyond the Factory Floor, see reference 1
  6. Thomas L. McGurk, “Ramping Up and Ensuring Supply Capability for Biopharmaceuticals,” BioPharm International, January 2004, pp. 1-4.
  7. Roger W. Hoerl and Ronald D. Snee, “Closing the Gap,” Quality Progress, May 2010, pp. 52-53.
  8. Ronald D. Snee and Roger W. Hoerl, “Further Explanation,” Quality Progress, December 2010, pp. 68-72.
  9. Ronald D. Snee and Roger W. Hoerl, “Proper Blending,” Quality Progress, June 2011, pp. 46-49.

© 2013 Roger W. Hoerl and Ronald D. Snee

Roger W. Hoerl is Brate-Peschel assistant professor of statistics at Union College in Schenectady, NY. He has a doctorate in applied statistics from the University of Delaware in Newark. Hoerl is an ASQ fellow, a recipient of the ASQ’s Shewhart Medal and Brumbaugh Award, and an academician in the International Academy for Quality.

Ronald D. Snee is president of Snee Associates LLC in Newark, DE. He has a doctorate in applied and mathematical statistics from Rutgers University in New Brunswick, NJ. Snee has received ASQ’s Shewhart and Grant Medals. He is an ASQ fellow and an academician in the International Academy for Quality.

Average Rating


Out of 0 Ratings
Rate this article

Add Comments

View comments
Comments FAQ

Featured advertisers