This past week I was working with a team on a project looking at some issues that were being seen in our overall surface transportation network. In this particular consolidation center, the data showed a lot of things going wrong. We were sent to "fix" things. Instead of just going in and changing things, we utilized the methods we have been learning through our Operational Excellence transformation and huddled together, then did our Gemba Walks. Following the Gemba Walks we came back to reflect together on what was going well with what we were seeing, and what could be potential obstacles in the process. This was mainly a learning week, as it takes a lot of effort to see a 24-hour operation. We spend long hours at the gemba, asking the people who do the work questions to get a better understanding of what they see as their biggest obstacles. I want to reflect a little bit this week on something we learned about the system as a whole, and the potential root cause to why this happening - individual performance reviews that reward heroism.
The quote above comes from Out of the Crisis by W. Edwards Deming. I've recently re-read and re-watched a lot of Deming's works. This particular quote is interesting to me this week. Imagine you are a manager of a large processing plant. It doesn't quite matter what you process, just know that you are one plant of many that interacts with all others. You were brought into the plant as a "fixer." Upper management has been watching the metrics for each individual plant and saw that your previous plant was doing so well that they decided you could go into any plant and fix it - you are greatly rewarded for this. You were then supplanted into the underachieving processing plant (7th manager in 3 years) to fix it. All those other managers that upper management tried out just weren't up to the task, but you are! You are the one to do it this time! Upper management tells you how great you are and they give you all the authority (and responsibility) for this other plant. First action they want you to take is to make sure all the product gets out of the plant and preferably on time.
Previous managers have tried different tactics, each one came in and tried to put his or her mark on the plant, making processing decisions that were "data-driven" trying to impress upper management with their quick decision making skills. Operations managers would hear "the data tells us we should do this." Looking around the operation itself, the operations managers didn't know what data the plant manager was looking at, but they're the boss! So now, you come in as a new plant manager, and you are feeling good about yourself. In your last plant there were very little problems, you felt as if the team functioned well, and it was all due to your leadership. You spent little time in the operation, you felt that the team could handle that on their own. Your job had been to look at the data and let everyone know what decisions the data was telling you to do. You decide to do this again. The operations managers have heard this all before. You decide, unilaterally, to hold all dispatches until all the product has made it on the truck. You are willing to hold trucks up to an hour after their scheduled departure. Surely this makes sense. What's an hour anyway? If you can wait an hour and get 20% more product on the truck then your performance scores ought to follow!
Weeks go by and scores fall. They fall dramatically. So much so that upper management decides, quite reasonably, that you are just like all the others, incapable of dealing with a complicated plant. If only there was someone who could help! Well, upper management seems to know a lot, so they send out a contingent of their top minds to take over the entire region. It's best to let the experts handle this. They look at the processing centers in the region and decide it must be a capacity issue. Just ship volumes between all the different centers based on forecast volumes each day. That ought to fix it, use the capacity of each processing center. Again, weeks go by and the scores only reflect worse and worse performance. Management from the top of the company now takes notice. That region's upper management just don't know what they are doing. It must be chaos in the processing plants for it to be so bad, utter chaos. Might as well send even more people to solve this problem. So more heroes arrive.
As the "fixer" plant manager, you realize quickly how little true authority you had, but surely you will be held responsible for these failures! No raise, in fact, you might just have to be sent back down a level due to your obvious incompetence.
While this little story seems exaggerated, I can assure you these things happen all the time. What was missing in the way everyone was thinking? They were focusing on the metrics, they let the data tell them what to do, so why weren't things getting better? Turns out, there's a very reasonable explanation for all of it, they just needed to spend some time at the operation to listen. Truck drivers, as it turns out, have a limit on the number of hours they are allowed to be on duty. Your initial decision to hold trips up to an hour ensured that the entire truck load of product would end up sitting on the side of the highway as the truck driver took his required rest. You didn't see this part of the system, but the truck drivers had been complaining to the supervisors of this issue, but you didn't listen - the data was telling you what to do. Then as more "expert" teams came in, so too did they forget about all the other parts of the system. You only have so many drivers to transport product between processing centers. Very quickly everyone started missing their deliveries.
This is something I've been observing in the past week. Many decision makers not seeing the operation from the gemba, and many more not understanding the system as a whole. As part of this team, we have a different mindset around what to do - we must learn. We were sent as "fixers" but we quickly adjusted to become "learners." In just a few days from being at the gemba, we saw disciplined processes that were well organized and had good flow (internally) - not much chaos in their own processes. The people working all knew what needed to be done, communicated well with each other, and you saw the results of that at the place of work - well organized and relatively smooth flow. What was more apparent is how quickly they could spot where the problems existed. They talked with the drivers each day and knew there were issues with the required rest they had to take, they knew there were too few local drivers to transport product multiple times between centers and still fulfill their required deliveries, and they knew there were constant late deliveries from certain other processing centers that would show as failures for themselves.
To visualize this, we created a value stream map. Instead of the typical value stream map that captured cycle times, we decided to try to visualize how variation within each part of the system might be causing variation throughout the system. We looked at delivery times of all input processing centers and created control charts for each. In just this step we saw unstable processes, with certain centers having upper control limits at over 4 hours late. Looking at the output control charts, deliveries from our processing center out to many others, we found that variation was quite reduced for some time, until of course a decision was made to hold deliveries up to an hour until all product arrived (mind you, it was not uncommon to be much more than 1 hour late). Still, actual products missed delivery times before the decision, and now even more after that decision.
We are taking each step in the value stream and using the data to help us understand the health of the process, its predictability. This leads us to go to gemba to see the process and get a better understanding of it. It isn't until we truly learn that we can improve. It isn't until we can figure out what is causing the instability, that we can make the process stable. Thinking back to Deming's funnel experiment, I imagine this is what is occurring right now within management. We see a point that is off, so we move the funnel - in this case, we move volumes from one center to the next, we change how we process. There is no constancy, there is no stability in volume, so how can we stabilize?
Why are we doing this to ourselves? When I speak with managers I hear the words as clear as glass - "I was sent to make them better." We have bred heroes who, due to lack of systems thinking, are doing more damage than good. The villain - variation and waste - is growing stronger every time they make a move. Perhaps it's time we take a step back, and instead of action, we gain knowledge, we learn. We have got to learn the system and understand variation so we can make actual system level improvements. Otherwise we create more failure, in the hopes that we will be rewarded for our heroism.