Shifting the Paradigm: 30,000-Foot-Level Metrics Provide Predictive Scorecards

In a previous blog, "Enhancing process capability reporting: An alternative to Cp, Cpk, Pp, Ppk process capability indices," a methodology was described for quantifying a stable process’ performance relative to its specification limits as an easy-to-understand nonconformance rate.

One application of the described 30,000-foot-level report-out is to quantify the response of an individual continuous-response process, which could become a baseline from which the benefits of improvement efforts are assessed. Another application of this reporting methodology is that the described technique can also provide the foundation for scorecard report-outs in an overall business management system, which will be illustrated below.

Performances reporting at the 30,000-foot-level needs to first determine if the process is stable. This can be accomplished using an individuals control chart, where the frequency of sampling is selected so that common-cause input variability occurs between samples. For example, if we were assessing hold time in a call center, we might expect that certain hours of a day would regularly experience longer hold times than other hours of a day because demand differences. Since we don’t want to react to hour-of-the-day differences as though they are special cause, one might then select an infrequent sampling plan that randomly selects one call a day to track at a 30,000-foot-level measurement.

Realize that the scenario described, in the prior paragraph, is a method of sampling and charting that is intended to monitor the process as it is seen by a customer. A method to monitor the management of the call center process would be created differently. If the intent was to measure the process in order to adjust and manage the process, in a Shewhart-control-charting view, then this would not be the most appropriate sampling method. If the goal is to manage the process, a better sampling plan would be to sample calls more frequently and then adjust the staffing at the call center as needed to keep the wait times at a target level. This could be called 50-foot-level process control, which is not what you would monitor for an executive scorecard.

Over time processes can experience a significant change from one level of stability to another. When this occurs, the 30,000-foot-level individuals control chart would be staged, where process capability statements can be made for each stable staged area. If a process has a recent region of stability, it can be said that the process is predictable. When this occurs, data from the latest region of stability can be considered a random sample of the future and used to provide a prediction statement.

This process exhibits a shift around day 23, which creates a need for a stage to separate the two different levels of performance. Since the process has a recent region of stability, the process is considered predictable and the data in that stage can be used to estimate the process’ after-change capability.

For this set of data, a 30,000-foot-level performance report-out would take the following format, where the text box below the plots provides the recent region of stability non-conformance rate (0.629 + (100-87.768) = 12.861) in terms that everyone can easily understand. Note how this report-out reflects a stage at day 24, where a new process was incorporated.

This form of predictive-process reporting can be applied to business scorecards through an organization’s Integrated Enterprise Excellence (IEE) value chain. An IEE value chain describes at a high-level what the organization does and the measurement of each function relative to a quality, cost, and time assessment. A full description of the IEE Value Chain will be included in a future blog.

Reference

The content of this blog was taken from Chapter 12 of IEE Volume III

Chapter 7 of IEE Volume II