How to Think About the Six Sigma Tools
The statistical and process improvement tools are clearly an integral part of the Six Sigma approach. As illustrated in this case study, without these powerful technical tools, Six Sigma would regress into a collection of vague concepts, slogans, and other fluff. It is the technical toolset that brings rigor to Six Sigma. The MBBs, Black Belts, and Green Belts are the resources primarily responsible for use of the tools. This raises the question, “How should leaders, such as executives, business unit leaders, and Champions think about the Six Sigma tools?”
First, it is important to understand that these leaders do not need to become professional statisticians, or even highly skilled in use of the tools. Of course, one can never have too much knowledge, but forcing leaders to become experts (as some executives have attempted to do) is a misguided approach. They simply don't need this expertise to do their job. After all, in the vast majority of projects, such as the newspaper case study, these leaders will not be the ones directly applying the tools. On the other hand, we do recommend that leaders study and learn the tools well enough to complete good Leadership Green Belt projects, as discussed in Chapter 6. During Green Belt training, leaders should actually complete their own projects, rather than relying on MBBs or Black Belts in their organizations to complete their projects for them. There is no substitute for personal experience with the tools and the overall DMAIC process.
Two extreme viewpoints that leaders should avoid are feeling they need to become experts in the tools, and refusing to study the tools at all. For example, when the topic of the technical tools comes up, some leaders will joke: “I was never good at math!” Such an attitude belittles the importance of the tools, and diminishes this leader's ability to hold meaningful project reviews.
The middle ground is for leaders to study and struggle with the tools like everyone else in the organization, and to apply them to their own projects. They should understand what the key tools are, when they should be applied, and what information each produces. This understanding will help them hold useful project and overall initiative reviews.
Leaders don't necessarily need to understand the mechanics of how each tool works, or the underlying mathematics. Rather, they need to understand how the tools fit in with the larger picture of the overall Six Sigma deployment. For example, they should understand and be able to articulate the following points:
The statistical and process improvement tools provide the rigor in Six Sigma (as noted earlier).
The tools themselves don't make improvements; rather it is the action taken by people based on application of the tools that generates improvement.
The tools must be properly sequenced and integrated in a disciplined fashion to be effective.
If the leadership aspects of Six Sigma are not in place, the tools will not have any lasting effect.
The tools should be combined with subject matter knowledge in an iterative fashion of generating, testing, and revising hypotheses.
The Tools Themselves Don't Make Improvements
Relative to the second bullet item, notice from the case study that all improvements to the publishing process came about from actions taken by people, not from the tools. For example, implementing (and actually using) well-designed job aids significantly improved accuracy. Of course, use of the tools helped identify the root causes of inaccuracy, and determine the best countermeasures to deal with these root causes. Pareto analysis was a key tool on this project. As with most Black Belt projects, rigorously defining the appropriate metrics to be improved (CTQs), and obtaining relevant data for improvement purposes were key challenges. In this case, developing a precise definition of an accuracy error was critical for progress.
The key point here is that the tools help identify the root causes of problems and potential solutions, but for improvement to occur people have to take action based on the tools. For example, numerous investment organizations lost large sums of money when the energy conglomerate Enron collapsed in late 2001. Interestingly, many of these investment companies had tools designed to predict such corporate defaults. The typical reason many still lost money was not some inadequacy of the tools themselves, but rather the lack of prompt action based on the tools.
The Tools Must Be Properly Sequenced and Integrated
The tools must also be properly sequenced and integrated in order to be effective. Figure 8-7 illustrates how the most commonly used tools are typically sequenced and integrated by the DMAIC process during the course of Black Belt and Green Belt projects. There is a logical progression, with the input of one tool often being the output from a previously used tool.
Figure 8-7. Six Sigma Tool Linkage
The process map, or flowchart, is typically the first tool used. The process map sets the stage for subsequent tools by carefully documenting a common view of the process. This map enables people to see their piece of the process in the context of the bigger picture. The cause and effect (C&E) matrix naturally follows the process map. Once the team agrees on the major steps in the process, it is logical to determine which steps and process variables are most critical to achieving our CTQs. The cause and effect matrix does just this, by noting how strongly each process step and variable impacts each CTQ. The process map and cause and effect matrix also provide input to the control plan by documenting the process steps and variables that need to be included.
Once the cause and effect matrix identifies the priority steps, the team will need to ensure that it can accurately measure the key variables at these steps, utilizing a measurement system analysis (MSA). In addition, it may begin a formal FMEA to identify potential failures in the prioritized steps and variables, and begin proactive counter-measures to prevent them. Once the team is convinced that it can accurately measure the key variables, it will likely evaluate process capability using capability analysis tools. Assuming the capability is insufficient, the team can use a multi-vari analysis to identify the key process variables that are causing the bulk of the variation in the process outputs (CTQs). Formal design of experiments (DOE) provides additional power to resolve ambiguities and quantify cause-effect relationships.
A key characteristic of the Six Sigma methodology is that the output of each of these tools provides input to the control plan, by determining the most important aspects of the process that need to be controlled to maintain improvements. This approach greatly simplifies development of the control plan because much of the hard work has already been done. Statistical process control (SPC), the other commonly used tool depicted in Figure 8-7, is then utilized by the control plan to quickly identify abnormal behavior in the process so that root causes can be found. Statistical process control uses control charts to document the range of normal behavior in the process, allowing early detection of potential problems before they become major issues.
This logical integration of the tools into an overall improvement process is a major contribution of the Six Sigma methodology. Statistical and process improvement tools have been around for a long time, and have been promoted by many other initiatives, such as TQM. However, instructors have generally taught practitioners a collection of tools without providing guidance on how to properly integrate or sequence them to solve a real problem. This approach often left people confused as to how to start, or where to go next after applying one tool.
Leadership Is Still Required
We have attempted to stress throughout this book that leadership is the key to Six Sigma success. This is also true for effective use of the tools. If the leadership component is lacking, no amount of tool usage can overcome this deficiency. For example, when the job aids were originally rolled out to the newspaper organization, subsequent data revealed no improvement. Why? Because people did not think management was serious, and so they did not use the job aids. If this situation had been allowed to continue, the project would have ended in failure. Fortunately, the editor exerted leadership by directly addressing the problem, and insisting that the job aids be utilized. Errors decreased dramatically in the next few months. There is no substitute for leadership.
Leaders should be continuously looking for similar situations in their own organizations where effectiveness of tool applications is being hampered by leadership issues (bureaucracy, politics, lack of clear direction, etc.). These issues tend to be the ones the Black Belts and even MBBs are unable to address on their own, and they are one reason for having a formal Champion role. Effective use of tools should be another area of focus at both project and overall initiative reviews.
Incorporate Subject Matter Knowledge
Keep in mind that the tools work best when they are combined with good subject matter knowledge in the technical area of the project, whether it is engineering, finance, or marketing. This is why it is helpful to have Black Belts who are knowledgeable in the areas in which they are doing projects.
Subject matter knowledge provides the theory to guide initial use of the tools. The information gained from the tools then helps refine, augment, or revise the original theories. This sequence continues from phase to phase, and tool to tool, resulting in greater and greater process knowledge. This is essentially application of the scientific method to business problems. A more detailed discussion of the proper integration of data-based tools and subject matter knowledge can be found in Hoerl and Snee (2002).
In the newspaper accuracy case, it was necessary to have the editor, copy editors, graphics editors, a reporter, and supervisors on the team in order to cover all the key areas of subject matter knowledge. If such subject matter knowledge were not critical, you wouldn't need such diverse teams. For example, it would likely have been impossible to determine the costs of finding errors at various stages of the publication process without such expertise. Subject matter knowledge also helped guide the original data collection, and helped interpret the results of the data analysis. Of course, the data analysis revised the team's original theories, dispelling for example the theory that the number of absent employees was a key contributor to the accuracy problem. As is often the case, analyses of data and subject matter theory enhanced one another when properly integrated.
Leaders should ask questions about what new knowledge the organization has learned from data analyses, and how current theories about the process need to be revised based on them. In short, we should have data to validate all our theories, and theories to properly interpret all our data.
In this arena there are again extreme positions that can inhibit project success. For example, some Black Belts may become enamored of the statistical tools and feel that lots of data and analyses are a replacement for process knowledge. Such Black Belts will tend to skip over the step of interpreting the data in light of subject matter theory, resulting in invalid conclusions. Conversely, some purists may believe that all problems can be solved based on first principles (fundamental laws) of engineering, physics, or finance, and resist any data collection and analysis. Leaders must push back on both extremes, and insist on proper integration of subject matter knowledge with data analysis.
If leaders understand these key concepts about the Six Sigma tools, and can properly audit for them during project and overall initiative reviews, they will add significant value to their organization's deployment. Conversely, those who try to become professional statisticians, or totally avoid discussion of the tools by using the excuse “I was never very good at math” will significantly diminish their value-added.