Seeing a book announcement for Duke Okes Performance Metrics: The Levers for Process Improvement made me think. The except titled “Psychological Impact of Metrics” included the notion that metrics drive behavior. That isn’t new, yet Duke went on to provide a way to determine the worthiness of metrics.
Here’s a short except from the except (see full except and learn more about Duke’s book here.)
Another useful method is to do a congruence check. This involves asking three questions:
• What is being measured by the organization?
• What does management ask/talk about?
• For what are people rewarded?
Note that if there is good alignment of the answers to these questions, there’s a good chance the metrics are the right ones; however, if there are significant differences between them then people are not being presented with a congruent environment— one that allows healthy discussions about how well things are working and what actions are necessary or useful to improve performance.
I’ve seen way to many organizations that list reliability as a ‘cornerstone’ of their products. And, then consistently talk about and emphasis reducing cost or time to market. To bring ‘reliability’ to mean more than platitudes the entire organization has to have a consistent message concerning the importance of product reliability. We need to use a metric that has meaning, is useful for decision making and is well understood.
As reliability professionals we can provide clear measures. MTBF is not one of those ‘clear’ measures. We can coach senior management on how to talk and ask meaningful questions. And, we can highlight the value of product reliability to our customers and the bottom line.
Once the organization measure reliability, talks about it, and rewards improved reliability, then we are using metrics related to product reliability hat are useful. There is a lot more in Dukes book so I’m looking forward to getting a copy to add to my professional reading list for the near future.
Great post. The psychology of metric selection and reporting is crucial for performance enhancement. Choosing a metric but not publishing the figures is useless. To change behavior one must select the right metric to drive performance in the direction desired. The metric must then be fully reported with recognition given to good behavior that produces the higher value metric. Collecting metric statistics without communicating the results will most likely lead to stagnant or even reduced performance. OUCH.
Thanks Barry and great comments. One thing I didn’t mention is those situations where the ‘hero’ that works all weekend to fix a design error – maybe even created by him – gets the bonus, praise and rewards. Then if a product goes out and meets or exceeds the reliability goals – nothing… that is frustrating and easy to fix with aware managers.
cheers,
Fred