I would like to start this discussion with a methodology and approach that we often use – that is the DMAIC approach of the Six Sigma Methodology.
DMAIC stands for Define, Measure, Analyze, Improve and Control. In today’s discussion we will cover what are in my mind the two most critical tools used in Six Sigma. These two toolsets are used primarily in the Analyze and Control phases of the DMAIC cycle.
To elaborate further on what lies under the covers of DMAIC, there are several steps or tollgates as demonstrated that are undertaken in each phase of a project to address improvement of an already established process.
In the Analyze phase there are three important processes conducted. We examine the data, watch the process, and determine the root causes using Statistical Analysis. In the Control phase we choose the right control method using Statistical Process Control and document the response plan.
Each tollgate addresses specific areas that –
- provide us the ability to define who are the customers and what are their needs,
- measure how the process is defined and determine measurement systems,
- analyze what are the important causes of process variation,
- improve processes by eliminating variation and non-standardization, and
- control and plan what actions are needed for stability, capability and continuous improvement.
The image that comes to mind when Six Sigma is mentioned is the bell-shaped curve. For those of us who have had the pleasure or pain of taking statistics in high college, these images should be buried in our memory banks somewhere.
First and foremost statistical analysis is a way of using statistical methods to analyze and present information about what is going on either directly within a process or how individuals and/or groups are performing relative to some measure.
If a process is normal and stable, data associated with it can be plotted. It has predictable characteristics that are typically represented as a bell-shaped curve. If we are operating processes at a 2-Sigma level, then our yield or throughput of operating is approximately 69%, which means 69% of the process is being done defect free. At this sigma level, we are creating approximately 309,000 defects per million opportunities (DPMOs). Therefore, using statistical analysis we can measure how our processes are performing relative to sigma levels.
For example, a call center operation has a number of people making service calls to clients. If they are operating at a Six Sigma (optimum) level, based on call volumes per day, week, or month, we know that their operational effectiveness as measured by the yield or throughput is 99.73%, and that they are creating only 3.4 defects per million opportunities.
This may mean that our customer service reps are following up on customer complaints, if technical support can be benchmarked. What is a defect in this case? Well, it can be not closing out the call within a specified time limit or not getting a customer rating above exceptional for the call. The defect is based on the criteria(s) (KPI) that are set by the company.
The table below shows the details of each sigma level at increments of 0.5-Sigma with their corresponding DPMOs (defects per million opportunities) and yields.
You may ask why is this important or where does my or should my company fit. This is determined by several factors:
- what your customer is willing to pay for levels of quality for the product you produce,
- how are your competitors doing, and
- what is your industry doing.
Certainly, it is not good enough for airlines to be operating at 6-Sigma. If so, it would mean 3.4 crashes for every million flights. The implications would be catastrophic for the airline industry and, more importantly, the customers. If we apply this same measurement to a heart surgeon, there can also be critical outcomes. He or she certainly does not have as much tolerance for defects as a claims adjustor working at an insurance company.
I hope that you get my point.