When people first learn of standard work within the Deming context (or lean manufacturing or the other flavors of management with some process focus) they often fear it means following outdated processes that are ineffective. I believe this is because, if they have experience with standards, that experience was a bad experience. They likely suffered from the equivalent of processes written in stone that no longer make sense but no one knew how, or no one had the authority to officially change them.
Often this results in people unofficially “changing them” (leaving them as they are, but) using their own variations. This often results in chaos (though often delayed until those knowing the right undocumented processes to follow are not around), including making process improvement difficult as the baseline state is so confusing.
One sign of progress in management systems over the last 20 years is there are many more organizations providing good experiences with standardized work processes (often from Deming based or lean manufacturing based management systems).
For standard work processes to be effective the organization needs to create a culture of process thinking (without that people don’t respect the importance of standard work and chaos and poor results ensue). Process thinking is a natural result of the Deming’s management ideas: evidence based management, PDSA, understanding variation, eliminating non-value added work, encouraging joy in work (which requires letting people view how they contribute to providing customer value), etc..
Randy Harward spoke at the 2013 Deming Institute annual conference on applying Deming management methods and sustainability and Patagonia.
It was better because designers and developers went from being just people who bought things, and marketed them, to people who had to understand and solve problems throughout the whole process. They became process engineers, every one of them… It was a huge change for the company. Sales and satisfaction increased for the whole company.
Previously, that that were thought impossible, technical barrier, just fell away as we explored the whole process.
Randy discusses how end to end process thinking, systems thinking, created the ability for Patagonia to create sustainability and business success together. Without looking at the whole system, sustainability often can seem to create increased costs. But as the entire system is optimized (for sustainability, quality and customer value) there are big gains for the business.
The culture of process thinking and continual improvement don’t sound much different than what companies say they do. But the difference in practice is profound.
Randy Harward spoke at the 2013 Deming Institute annual conference on applying Deming management methods and sustainability and Patagonia.
I was hired to take care of this huge volume of returns coming back from customers, we had so many quality problems. I was managing customer service and all the systems required to handle this huge volume of returns. And I kept raising my hand and saying “why don’t we do something about the problem.”
I visited companies, suppliers and customers all over the world. And I took seminars, and that is when I discovered Dr. Deming and it changed my life. It aligned much of my thinking.
I finally realized what was happening in a Western Mountaineering situation wasn’t… You know at the time at Western Mountaineering I didn’t think what was real management, I thought we were just goofing around. I have actually tried the rest of my life to recreate that situation… it is very very difficult to do what happened there.
We set to work building and documenting every system for design, development, production and sourcing of apparel worldwide. Part of that was teaching a 6 week quality course on tools and Deming’s philosophy to every employee at Patagonia… We taught the red bead experiment as part of that, and I tell you to this day I can’t believe what an impact that exercise has. It really changes world views. 25 years later people still tell me about that game.
What it [QA manual] created was clarity. What we had was photographs, detailed photographs of every stitch and type of construction. We started removing all the debate about what was and wasn’t acceptable.
When it was implemented it was like turning off the facet, seconds [bad product] just disappeared – at least half of them overnight.
Visual management and clear expectation are an important part of success. They integrate with the idea of standardization to make clear what is expected. Standardization doesn’t not stifle improvement, standards are meant to be continually updated. If the system doesn’t have a good process for continual updating the standards and visual job instructions that is a serious failure.
Randy Harward spoke at the 2013 Deming Institute annual conference on applying Deming management methods at Western Mountaineering and Patagonia.
Western Mountaineering was facing a cash flow crisis when Randy Harward took over and he strongly considered closing down. Given the problems the “only way out of the situation was to increase turns” (increase how quickly they sold their existing inventory). Then they could pay down the debt using the marginal profit from sales. Given the financial crisis, the group took up to 20% voluntary pay cuts.
Randy had all employees attend weekly finance meetings where they discussed the company books and the reality they faced (debt levels, importance of quickly turning inventory, cash flow, etc.). On top of this each clerk was given a budget to work with for their area.
each of them got to work on improving turns and sales in their area and the place started to get an energy it just didn’t have before. Everyone was involved at a level I have never seen and sales started to increase.
They had success but then ran into a new problem: they ran out of inventory. And when Randy asked suppliers to provide more inventory they were unwilling to do so (presumably due to concerns about being paid for their good in the event Western Mountaineering ran into financial problems).
In discussions about goals, I typically find attempts to create two distinct categories of goals. I see the words “arbitrary goals.” Arbitrary numerical goals are believed to be bad, problematic. Some numerical goals the non-arbitrary type are believed to be useful, good, even necessary. I could find no evidence Deming made a distinction between arbitrary and non-arbitrary goals. His distinction was between goals I set for myself and numerical goals others imposed upon me or set for me.
For me any goal is inherently arbitrary particularly a numerical goal. Deming seems to warn us that the ill effects of a numerical goal have to do with imposing it on others (without providing resources to achieve it) rather than whether or not the goal is arbitrary. Deming pointed out the damaging effects of setting numerical goals for others without distinguishing between arbitrary and non-arbitrary goals.
Thoughts on “Arbitrary”
When I establish a system boundary, it is a thought process. The boundary I establish is arbitrary. The system boundary is not in the world outside of me; it is in my thinking. The boundary is arbitrary not in the sense that it is without foundation or irrational, or whimsical or frivolous.
It is arbitrary in the sense that it is discretionary, voluntary, elective, optional, and subjective. I can engage in much thought and effort; utilize my current knowledge; collect and examine data and information as a precursor to setting the boundary. Yet even with all that effort the boundary is arbitrary in the sense I have described. It is based upon the subjective things like: my purpose in establishing it; my perception of my current context; my current knowledge (which is never complete); the data and information I elect to gather and the data examination option I choose to apply. The system boundary is not in the world outside of me; it is in my thinking.
Similarly, a numerical goal, regardless of the level of rational thought and amount of effort I invest in establishing it, is similarly arbitrary.
Deming’s use of “goals” and “arbitrary”
In a comprehensive search of The New Economics 2nd Edition, I found:
“arbitrary” was used only once [p 146 in regards to student assessment scales]
“goal” paired with “numerical” was used a dozen or maybe two dozen times
Support of top management is not sufficient. It is not enough that top management comity themselves for life to quality and productivity. They must know what it is that they are committed to — that is, what they must do. These obligations can not be delegated. Support is not enough: action is required.
As discussed in our previous post: Improvement is a Learning Process. If you are not practicing improvement yourself, learning is much less than it could be. Actively experimenting with improvement yourself maximizes what you learn.
Since executives have the greatest authority, they have enormous influence on the organization. If executives don’t learn to understand variation, continually improve using the PDSA cycle, appreciate the interactions within a management system, etc. they will stymie efforts to improve the management system.
Those with the most influence should have the best grasp of management improvement. They need to continually be learning how to improve the improvement process. And many of the processes they are involved with have the largest impact on the future success of the organization. Those are precisely the processes most in need of continual improvement.
In this segment of Ian Bradbury’s talk at the 2013 W. Edwards Deming Institute conference he discusses variation and using the control chart to aid improvement efforts.
This last point went outside the control limits so it is a signal of special cause. For that particular point it makes sense to ask the question – what was special, what was different about the conditions underlying the generation of this point of data versus the others.
If we didn’t have something like that, if all of the points were varying inside the control limits then it doesn’t make any sense, or it is unlikely to be a productive exercise to be asking why one point is lower or higher than another. That doesn’t mean there isn’t anything to learn, there isn’t anything to improve a system like that… to the contrary the improvement process is one where first of all we focus on the special causes of variation… and get the process into a state of statistical control.
Once we get it into a state of statistical control, then we move into a different kind of question. Now the questions is, what is common to all the underlying conditions that generated all of these points? What is the underlying system that generated this? What kind of system changes might we make that could cause a change in the overall variability of these results?
Understanding how to use control charts to manage more effectively is one of the areas that normally takes people a bit of time to become comfortable with. We are so used to treating every data point as special it takes awhile to improve our ability to learn by using special cause thinking (where we seek to learn why that point was different) only when it is the most effective way to learn (when their is evidence a special cause exists) and otherwise (most of the time) use common cause thinking. When the point is just the natural result of the process which includes variation from one point to the next it is not effective to think about what is special about a specific point, the more effective method to improve is to improve the process that generates all the results.
The improvement thinking that is most effective when there is no indication of a special cause is one which looks at the entire process and uses experiments (the PDSA cycle) to try out process changes. Based on what is learned from the experiments, then change the process to incorporate methods that improve results.
Using control charts effectively really isn’t very complicated; though it is something that does seem a bit difficult for people to learn to apply on their own. It is one of the areas where having a coach or consultant help at the beginning is often more effective than just trying to read about it and then apply it yourself.
Though really it is no more complicated than what is stated above: use our normal way of thinking (treating the data point as a special cause) when a special cause indication is shown and use the PDSA improvement/learning cycle when there is no indication of a special cause. Of course, actually doing that isn’t quite as easy as reading it or just understanding the implications of managing that way.
In this segment of Ian Bradbury’s talk at the 2013 W. Edwards Deming Institute conference he looks at how your view of the system effects what solutions seem reasonable. If you only view part of the system the solutions you come to can often have negative consequences.
And often those consequences are pushed off in time so that it can be very difficult to understand the causes of the consequences. Taking the time to train yourself to think about the larger systems that are in play and what impacts decisions may have on those larger systems will help you make better decisions.
How it is that you view the system can significantly impact your behavior. And [your view] can cause you to look at the system very differently, [for example] if I want to have sustainable fishing how do I need to interact with the system?
The initial results of actions taken may not have the largest impact – such as the example in Ian’s presentation where interaction affects create a much larger impact over the long term than the initial results. So that while there is a slight move in the beneficial direction at first once the system absorbs the change and rebalances the long term result is a negative one.
Evidence based thinking isn’t as easy as just looking at the results. We must have knowledge of the system that provides an ability to understand the results and comprehend that short term results may be misleading. As Dr. Deming said, “there is no substitute for knowledge” (chapter 6 of The Essential Deming provides more details on this thought).
The Lean Enterprise Institute has published webcasts of discussions with James Womack, Dan Jones and John Shook exploring reflections on 25 years of lean which I think are quite good. This excerpt looks at learning from other communities of interest and ideas used by the lean startup.
I agree with the idea that different management communities can learn from each other. And I think lean thinking and lean startup communities are two that mesh very well with those that believe in Deming’s ideas. It is true there are plenty of bad implementations of lean (usually with very superficial understanding of lean that are use some terms and maybe try some tools but not much else) but there is also a large group of lean folks that practice a lean thinking model that works well with Deming’s ideas.
It is articulated as a learning cycle. A learning cycle in which you are trying to build both knowledge of customer value and knowledge of the means of production for customer value [the performance of the process] simultaneously.
This may not immediately strike you as profound, but I believe it is. The focus on process thinking and performance as seen by what the customer values was not common even in 1990, or even today (though it is more common today). And in 1951 it was a very radical this view compared to how organizations actually operated.
The theory of knowledge teaches us that a statement, if it conveys knowledge predicts future outcome, with risk of being wrong, and that it fits without failure observations of the past.
Rational prediction requires theory and builds knowledge through systematic revision and extension of theory based on comparison of prediction with observation.
And the video includes the nice illustrative story of Chantecler the rooster (also detailed in the New Economics) to show how a theory is falsified by one failure of the theory to match reality.
In [the PDSA] cycle there is a synthesis of two things. There is a synthesis of taking action aid at improvement and learning. There are cycles that look similar, but they are not the same…
The PDCA cycle, as it is commonly described, is really an improvement cycle; not a learning cycle – if learning happens it happens by accident in the PDCA cycle…
There is also the scientific method – which is a learning cycle. It is a cycle by which you are testing hypotheses. But it is not an improvement cycle. It is not a cycle aimed at making changes that are an attempt to cause improvement…
The really critical piece of the [PDSA] cycle then is like we had in the story of Chantecler – prediction: prediction and the implicit theory upon which the prediction is based. So you are making change that is aimed at improvement but you are doing that at the same time as making predictions about outcomes articulating the theory. And then when outcomes are not as predicted… then we have cause to change the theory – cause to learn.
I really like this clip from Ian’s talk quite a bit (and the whole talk). Watch it and hopefully you will enjoy it also.