By John Hunter, author of the Curious Cat Management Improvement Blog.
Kevin Murphy’s presentation at the 14th In2:InThinking Annual Forum Weekend Conference: Application and Lessons of Deming’s Perspective on Leadership:
[Vimeo plays the video automatically – which is obviously an extremely bad Ux practice. Even using their code to disable autoplay of the video doesn’t work. So I have removed the embedded video. You may view the video here.]
Kevin is the President at Triumph Accessory Services. After graduating from the Deming Scholars program at Fordham University he worked for GE Aircraft Engines. When GE decided to sell off the units providing engine and airframe accessory services, Kevin participated in a management buyout of the company that is now part of Triumph Accessory Services.
The sound on the video is not great. It is less bad on the last half, so if you really can’t put up with it at the beginning you could skip to about half way through the presentation.
Kevin discussed the fact that many important factors cannot be managed and the importance of managing with consideration of those factors and impacts. He also showed the importance of using data when possible to help manage your organization. Sometimes I get the impression that some people learn the idea that some important factors cannot be measured and think that is an excuse to not worry about using data. This is not at all what Dr. Deming meant.
Data is extremely useful in managing an organization and doing so well is critical to continually improving. Dr. Deming’s admonition that “It is wrong to suppose that if you can’t measure it, you can’t manage it – a costly myth.” was a warning of trying to force the use of data into decisions where they didn’t help. It was not a statement against the many cases where using data is extremely valuable.
Don’t be too quick to use the excuse that some things can’t be measured when you are having difficulty figuring out how to measure to access the current status or the effectiveness of experiments. Do be willing to accept that some things cannot be measured, but in my experience it is far too common for people to give up too soon thinking of useful ways measures could be used. Kevin showed a great many examples using data and control charts to lead improvement at Triumph Accessory Services.
Kevin also discussed an example where the data begins to look troubling. And it shows the importance of knowing exactly what is going on with processes for which you are using data. The process was under experimentation and the extra effort in trying to find good solutions increased the time it took to complete the process.
If we had setup labor standards there is no way, no way that we would have gotten the experimentation and the improvement we got. There is not a chance.
He is using “labor standards” there to mean setting standards such as set times for specific tasks that people are measured against and evaluated on (the quote is from about 78m 45s).
When organizations are short term focused or have executives making decisions based on data without understanding the processes generating that data in detail they will “hold people accountable” for bad data and drive fear into any thoughts of experimentation. With a Deming mindset allowing those close to the process to experiment and accept short term costs Triumph Accessory Services was able to reduce the average cost, in this case, from $1,404 to $388 over a 4 year period. But it required trusting those doing the work to experiment wisely and not having executives react to initial cost increases and block improvement efforts.
I liked how Kevin explored the issues that crop up as you try to use data. It isn’t as simple as it sounds. Interpreting data accurately requires knowledge: both of how to understand data and the systems generating the data. He discussed, for example, Simpson’s Paradox which can lead you to draw incorrect conclusions. This happens when the data you are using as a proxy for measuring a process could look bad not because the underlying process is providing worse results but because of the mix of sub-grouping in the data changes (related thoughts: stratifying data and The Value of Displaying Data Well).
I actually was thinking earlier in his talk that the data might have looked worse (over time) because the mix of complicated servicing (that required more time) versus simpler servicing changed: the data he showed had a lot of variation which could be an indication of mixing 2, or several, fairly different processes into 1 chart. I didn’t know enough about the processes to know if that was a reasonable possibility but that is a question just looking at the data he was showing raised in my mind.
Kevin also discusses the problems with contingent incentives (awards, etc.). He eliminated such awards at Triumph Accessory Services (related posts: Eliminate Slogans, Exhortations and Targets, The Trouble with Incentives: Even if the Organization is Destroyed in the Process and What Really Motivates Us?).
That’s really what we are trying to do in transforming a company. Establish trust, that is going to lead to interest and focus. That is going to lead to innovation in the process… the long term survival of your company really depends on innovation. (quote: 81m 30s)
Related: Paula Marshall’s Presentation at Our 2015 Conference on Using Deming’s Ideas at The Bama Companies – Deming Podcast with Scott Dalgleish, CEO at Phase IV Engineering – Applying Deming’s Management Thinking at Patagonia