Six Sigma and big data has taken metrics to a new level. There they are measuring progress, pointing the way to improvement, and determining how big the bonus will be this year.
Knowing where you’ve been and where you’re going is a good thing, so what’s wrong with using metrics as much as possible? Beware of unintended consequences!
Humans are funny and usually not stupid when an objective is placed in front of them. If their performance is measured based on a set of measured objectives, don’t be surprised when you get more of that behavior. Well designed metrics can and do have the desired results. The real concern is if other things are neglected or even sabotaged to improve the desired performance objectives.
Suggestions for better metrics
The Easy to Measure Pitfall
Are you measuring something just because it’s easy to measure? Low importance data that are used as a substitute for things that are important, but too hard to measure should be avoided. A qualitative measure of a hard to measure metric is better than feeling good about a worthless, easy to measure data set.
Avoid Metrics That Don’t Tell the Whole Story
Metrics can leave out the whole story, making it difficult to understand the underlying issues.
Make Sure All Metrics Align with Your Real Objectives
Any performance measurement used should directly indicate how well your objectives are being met. The real metric may (likely) be a composite of two or more metrics. For example, let’s say we want to track labor cost per assembly. Labor cost is composed of assembly time and labor rate. If we just tracked assembly time or labor rate we wouldn’t be measuring what’s really important, the cost to assemble. We should probably track assembly time to monitor any variation, but tracking it alone will not tell the whole story. Assembly cost may vary because of labor rate going up due to factors such as increased overtime.
Beware of Unintended Consequences
This is where being creative comes in handy. Think of the ways that reporting a metric may alter behavior of the system, especially humans. Are the metrics altering behavior in a positive way to achieve desired results, or will meeting metric targets cause harm elsewhere?