HPS Leadership Best Practices Journal™

BPI-6 Sigma: The Measure Phase

Let's briefly review where we are. An operational issue is identified within the company. We have little knowledge of the underlying cause, but the issue has big impact upon the company's performance (quality, cycle time, cost, productivity). Executive staff and the Black Belt determine the issue to be a good BPI project and approve a BPI team creation to attack the issue. A contract is developed and signed off by vertical management indicating support to the project and team. The contract information is entered into the project database using the standardized contract format. The team is formed comprising employees familiar with the issue and who work in the area impacted (our "sandbox"), the first meeting is held, and various roles are assigned for the functioning of future meetings (similar to the dynamics of Total Quality Management; TQM roles are facilitator, time keeper, scribe, minute taker, etc.). A process map is developed of the "As Is" process, including decision points, queues, electronic/hardcopy inputs/outputs, and process notes. All of this happens in the Define Phase, the first module/phase of the BPI model.

The second phase of the BPI model is the Measure Phase.

Broad measurements (data) should reflect process performance relative to quality, cycle time, cost, or output (volume), consequently reflecting the voice of the customer. Remember, the customer is the entity that receives the output from the process. Since data collection consumes resources and time, it is important to collect only that which will be useful in depicting the process dynamics. Data tells us where we are and where we wish to go! The output measures reflect the customer's requirements and expectations. This requires the BPI team to understand who the customer is and their needs and requirements.

By starting with output measures, one can look upstream in the process to identify measures that support the output and that identify variability. Variability exists in all processes. We have variability in the process product and variability in our measurement system. Before any changes should be made to the process, you should attempt to squeeze or reduce the process variability and fine-tune the measurement variability. If you don't, then any changes you make in the process steps might affect the output only because of the natural variability that exists, and not because of the changes made. When looking upstream, our purpose is to identify those vital few input and process factors that affect the output. This is where we need to focus our attention. Remember that measures should reflect the process and not an individual's performance. Measurement provides a feedback regarding the process behavior, where fine-tuning is required, and where the issue needs to be analyzed further. Measurement is a key toward improving the process. I've heard it said, and I agree, that "An opinion without data, is just another opinion!" And there will be a lot of opinions regarding the intuitive fix.

The types of data that will be measured will be attribute data and variable (continuous) data. Attribute data is usually characterized by countable outcomes and requires a relatively large number of responses to be useful for analysis. On the other hand, variable data is more measurement in nature. This type of data is rich for statistical analysis. A variety of literature (some listed in the first article) is available to further define and describe the characteristics and types of statistical analysis that can be performed using these types of data. To facilitate the data analysis, statistical techniques involving sampling (design of experiments, sample size determination, sampling plans), inferential statistical analysis (confidence intervals and test of hypothesis) are frequently used. A systematic data collection process ensures continuity and provides ongoing monitoring of the process. Training sessions for the data collectors, well-designed data collection instruments, standardization of terminology and criteria, how the data reflects the customer requirements, quality control (QC) charts, statistical process control (SPC) gauge capability (how good is the measurement system), and other statistical methodologies provide the pulse of the process that is understudy.

In the next BPI model phase, Analyze Phase, we will look at various quality tools that assist us in analyzing the process and data obtained from the process.