The senior executive was both adamant and insistent. He wanted reasons. Now.
The question is valid. The answer is elusive. But NOT impossible.
Why DO the numbers go up and down? Why does the yield seem good one week and lousy the next? Why does the labor used for Process X change from day to day? Why do the same steps in the same work sequence yield differences in quality over time?
As in many cases, Dr. Deming offers guidance here.
His view of variation is captured in his description of common cause and special cause. How does this work?
In short, he urges users to plot the metric and calculate the control limits of the data series, generally two standard deviations above and below the mean. Inevitably, the data will move around. How to explain the variability??
Dr. Deming says movement within the control limits is due to common causes, the inherent variability in any system, the noise, the general changes which read on any process. How to improve it?? You work on the system, apply continuous improvement. There is not a one-to-one correspondence between cause and change, however, much to the chagrin of the questioning executive.
Conversely, movement beyond the control limits is due to special causes, one-time, traceable aberrations, both for good and bad. Unit costs suddenly improve? A one-time purchase of raw materials from a supplier who ran out of warehouse space. A rash of late deliveries? A broken, important piece of production equipment. How to improve it? Go to the source of these one-time issues and fix them. You can also get a smile from the executive with your crisp explanation.
Teaching about, understanding, charting and measuring the key variabilities in a system is a wonderful training tool for any organization. It forces correct conversations about important issues. It can turn opinionated arguments into data-driven solution-fests.
One of the most concise discussions of this tool is in Marypat Cooper's excellent Kaizen Sketchbook. I recommend it highly.
Keep on learning.