Successful Operational Leaders Get This Right

One of the best lessons given to me as a young operational leader was around the concept of variation. Understanding this concept allows a leader to know when and how to make improvements to their processes. I picked this up from Dr. W. Edwards Deming, a trailblazing early leader in the quality and management movement. He is also well known for teaching post-WWII Japan how to "build quality in" versus "inspecting it in." While his writings went counter to my early beliefs around management, they were thankfully absorbed and applied.

Dr. Deming said one of the most significant mistakes management makes is how they respond to performance variation. Specifically, confusing "common" versus "special" cause variation and their corresponding countermeasures.

At a very high level, common cause is the expected, usual, historical and predictable variation in any system. It’s where 98% of the “data points” fall if the system is “stable.” Special cause variation is when something material and different occurs, throwing a data point or points outside the expected norm. Common cause variation is made up of several factors always present in that system. Conversely, special cause variation is typically caused by one factor behaving very differently and suddenly.

You may be thinking, “this is only useful in manufacturing,” but variation exists in any process or system. Whether you are managing 911 service levels for inbound calls, car repair times, billing errors, or just driving to work…each one of these “systems” produce variation. Also, how you respond to these two types of variation is very different. Dr. Deming said, “confusing special causes with common causes will only make things worse.”

Let me illustrate using a story from one of Dr. Deming's many management "interventions." I was told this story by an operational mentor of mine years ago.

MOVING BEYOND "FIRE FIGHTING" TO DRIVE IMPROVEMENTS

Dr. Deming was once hired to help a small tire manufacturer improve their quality. They had a higher percentage of defects than their competitors, adding costs and squeezing margins. They produced a predictable level of defects each day, so it was a "stable" system, but not capable of generating the results they needed. In turn, the new General Manager of the factory was given the task of making the required changes to improve quality significantly. While this was a noble cause, the more the General Manager tried to reduce the defects the worse things became. This was when they called in Dr. Deming.

The defects in question were concerning the thickness of the tires, as they had to meet very tight specifications. Sometimes they were too thin or too thick, and if outside the specification tolerances were deemed defective and thus scrapped.

Dr. Deming started his inspection of the factory and immediately saw a number of tires leaning against a wall. He asked why they were there, and the General Manager said these were the defective tires. He further explained that when they produce a defective tire, they stop the machines, inspect the tire, and then re-calibrate the machines accordingly. This was a new “quality step" the General Manager put in place to reduce the defects.

Dr. Deming asked, “Is this new step reducing the number of defective tires?” The General Manager responded, “no, the introduction of this quality step came about the same time we started seeing more defects. But this new step makes so much common sense. It must be helping us, so we kept it in place.” The General Manager said the issue was probably with the afternoon teams, and maybe an “after-lunch” lack of focus or poorly trained employees.

It was explained to the General Manager that “tweaking” a predictable and stable system after each defect could indeed be making things worse. Dr. Deming asked that the systems and machines run for a few days, no adjustments for any defects. The General Manager shook his head in disbelief but agreed. The system ran for several days, a predictable number of defects was generated, but nothing in the system was adjusted.

They reviewed the data together. Nothing immediately stuck out. They then saw something with the timing of the defects. Evidence was found to suggest that the afternoon teams were the issue, but that still didn’t fully explain it. How could all the late shifts be bad? Also, the defects started to happen at the end of the early shifts. They decided to get even more detailed as to when the defects took place.

After some more digging, leveraging simple improvement techniques, they finally found the root cause issue. There were a few big windows in the facility's ceiling, and around 2 PM the sun would shine in and very slightly increase the temperature inside. This would change how the chemicals in the materials behaved, and the thickness of the tires would be impacted enough to fall outside specifications. In response, the system would be adjusted. Then the temperature would drop around 4 PM when the sun had passed. More defects and adjustments would then take place.   

They fixed the issue by changing the location of those ceiling windows while adding tighter controls to keep the facility's temperature constant. Almost immediately quality increased to record levels, and the performance variation from the system tightened. The employees were also happier as the “blame game” had stopped.

THE LESSONS TO BE LEARNED AND APPLIED     

There are several lessons with variation to be learned from this story. Let’s review them.

• Know the type of variation you have – Dr. Deming most likely used basic control charts to determine that this system was exhibiting common cause variation. It produced a stable, predictable outcome, what statisticians call an “in control” process. Control charts are very straightforward and easy to use tools, and can typically be run through a spreadsheet application like Excel. Any operational leader should have them for their key processes and sub-processes. This is especially true for those processes that produce mid to high-frequency outputs. Remember though, just because a process is stable does not mean it is performing well.

• Improvements to common cause variation require looking at the whole system – Because several embedded factors create common cause variation, you need to see a wide range of data to find the root cause(s). Making adjustments error by error works well for special cause variation, but can make things worse with common cause. Think of the General Manager reacting to every tire defect. You need to observe a stable system as a whole to better spot “cause and effect” patterns. For example, you may notice most errors happen within a specific time of the day/week/month/year, with a particular product, with a specific process or sub-process, etc. You cannot see these patterns easily if you are just reacting and deep diving each issue one by one.

• Special cause variation must be addressed quickly – While there should be a sense of urgency to improve quality regardless of variation type, special cause should seem like a “fire alarm.” Because something big and different is driving this variation, immediate attention needs to be given to find and correct the situation (or repeat it if a good thing). Typically root cause analysis is more straightforward for special cause as it’s more visible and attributable to one factor.

• Don’t immediately blame the people working in the system – Deming said that based on his experience, only 6% of the issues he saw with a system were tied to the people in it. However, blaming people is the typical starting point when improvements are being asked for. We saw that with the General Manager, blaming the performance on post-lunch focus issues and poorly trained workers. The Ishikawa tool (a cause and effect analysis technique) is very effective in overcoming this tendency, as it forces a discussion around all six levers that make up a system. Think of them as the “5 M’s and a P”…Measurements (reporting, analysis, goals/specifications), Methods (process, process steps), Materials (inputs), Machines (tools), Mother Nature (weather, temperature, etc.) and People.

• Invest in process improvement disciplines and tools – Get some training on Six Sigma and other quality disciplines. Work to build these skills with the people already on staff, and you will reap the benefits. I recommend getting started with a few fundamentals, small steps. Based on my experience, most of the ROI comes with a few essential tools, techniques, and the right mindset. Be careful here, as I've seen some organizations go overboard on "quality" disciplines. They lost sight that these disciplines were a means to an end, not the end itself.

• Validate that your process improvements are “real” – Leverage those same control charts mentioned earlier to validate the improvements you made really changed the system. If they did, you would see special cause variation. Many times, common cause fluctuations are incorrectly attributed to process improvements that were made around the same time. This is a big mistake.

One additional observation from my experience. If you ask "what happened" concerning any up or down in the metrics, most people will produce a definitive answer to that question (right or wrong). So be very careful about when and where you ask it. For special cause scenarios, "what happened" is still the right question. With common cause, you must change the question. For example, "how fast can we review the data and get to a root cause."

These concepts and lessons are especially useful to inexperienced, high potential operational leaders. This high potential talent will typically be assigned to extinguish performance “fires” throughout an organization. As they are successful, their "firefighting" techniques will become more rewarded and ingrained. They think just react quickly, put in a lot of hours, and the improvements will always come.

The issues surface when they are asked to lead large operational groups that are stable, but significant improvements are still needed. Those “firefighting” techniques will no longer work, and in some cases will make things worse. They will also confuse their teams, reacting to every up and down in the numbers. They must learn to correctly diagnose variation and use common cause improvement methods to succeed.

Clarity around how to identify and respond to variation will not by itself make someone a great operational leader, that, of course, requires so much more. Even Dr. Deming's teachings went well beyond this one gem. Still, I believe understanding how to leverage this concept is a load bearing “brick” in any operational leader's foundation.

Image by Florian Berger