Measuring Performance of a Process
Measuring Performance of a System:
Use of Averages or Process Flow Charts
1. As collection and analysis of data becomes easier “data collection” and so also “data exhaust” has increased. We at times are collecting data that is not needed or discarding data that was needed. This as per Donald Wheeler, the author of Understanding Variation – They Key to Managing Chaos, is due to a person being numerically naïve – having no understanding how to digest numbers and extract knowledge that maybe locked up inside the data.
2. Measuring a systems performance by comparing it to specifications like the average performance or last month’s or last year’s performance is not the correct approach to gauge the performance of a system. This is because some months the performance will always be above the average and some months it will be below, because we are comparing it to the center.
3. No comparisons between two production values can be meaningful unless the context of that comparison is known. The current production and previous productions are subject to variation, also known as the voice of the process or the noise in the process and it is difficult to ascertain as to how much of the difference between the past and present performance is due to this variation and how much of the difference is due to the real processes. Once we know as to how much the output is changing due to inputs and how much of it is changing due to usual system noise will we be able to make useful conclusions.
4. Every process has two kinds of variations, one is the routine variation, which is expected even when a process has not changed and the other variation is exceptional variation, outside the bounds of the routine and this is to be interpreted as a signal that a process has changed. A manager needs to separate out the variations between these two components to be able to meaningfully analyze his production data and see if a change is needed or everything is good. A way to do that is by using what was created by Walter Shewhart – the control chart or the process flow chart.
5. The control charts provide a means where the extent of the routine variation can be characterized between two limits, the upper control limit and the lower control limit. These limits give us an idea of the routine variation that occurs in a system due to its noise. A variation which is beyond these limits is the exceptional variation and is caused due to a change in the process. These changes need to be studies to see as to what is changing the performance and what can be done to improve it if it is going down.
6. Also, since the specifications like average production, yearly production or monthly production are not to be used to analyze a process does not mean that such specifications are not required. These specifications are facts – they tell you if the bottom line shows a profit or not, are the sales increasing as fast as the expenses or not and they can be used for planning the budget or making predictions, however these are not to be taken as targets. As per Donald Wheeler, predictions based on past data should be used for guidance and help and not for judgement and blame. However, care should be taken to ensure that specifications being used should be based on facts and not arbitrary numerical values or goals.
