The A in DMAIC is about finding the top two or three causes of the problem we are attacking.  When well executed, this phase starts with team input regarding potential causes, and uses statistical methods to isolate the top 2 or 3 causes.

Explore potential causes – listen to the team

Recruiting experts who are most familiar with the product and/or process  will drastically improve the team’s likelihood of a successful outcome.  Team members that work with the process on a daily basis can be extremely valuable as well.

There are a number of methods for exploring potential causes, and a Fishbone Diagram is a great way to categorize variation sources and develop a complete list. Additionally, team members can collect data between team meetings and help build Pareto charts.

Isolate the top 2 or 3 causes

This is the most important, and often the most challenging step in a Six Sigma project.  While the first half of the Analyze phase focuses on gathering team input, the second half must focus on eliminating the “trivial many” causes and identifying the “vital few.”  There are a number of tools to help accomplish this task.

In manufacturing environments, we’ve found the Shainin Methods to be extremely powerful in quickly isolating the vital-few causes, without interrupting production.  Techniques such as Paired Comparisons, Multi-Vari, and Components Search are among the most powerful that we have used.  We highly recommend Keki Bhote’s book, World Class Quality, which details the use of these tools.  Keki is a Motorola veteran who has solved some incredibly challenging problems in his time, and he is also a very motivating writer.  Also, the Shainin Group is one of the best consulting teams for manuacturing problem solving anywhere.  If you are a manufacturer with a  sizable training budget, give them some serious consideration.

The analyze phase cannot conclude until the top causes have been confirmed.  Among the most useful tools we’ve seen in this area are the Shainin Methods,  hypothesis testing, regression analysis, and basic design of experiments (DOE).