Why Six Sigma Failed
Why Six Sigma Failed
“There are three kinds of lies: lies, damned lies, and statistics.” -Benjamin Disraeli
“In God, we trust; all others bring data.” -NASA 1984
The following is an excerpt from my book that didn’t make the final cut.
As you learn more about Getting Shit Done (GSD), a thought might be, “Great, yet another productivity framework — I think I’ll pass; next.”
That’s why we’re taking a quick detour here to compare GSD against Six Sigma, which was perhaps one of the most overhyped management fads in recent memory. There are valuable lessons to be learned from what Six Sigma got right and what Six Sigma got wrong to aid our cause.
If you’re not familiar with it, Six Sigma first appeared on the US scene during the late 1980s and was promised to be the panacea for all the manufacturing and process improvement woes. The concept was initially adapted from Japanese business practices, Americanized then repackaged and adopted by some of the largest US manufacturing organizations like GE, Honeywell, Motorola, 3M, and DuPont.
Statistics ❤
Six Sigma’s overall premise is that a company can run much more efficiently by removing waste and almost all defects from nearly all processes by using statistics. By definition, Six Sigma stands for eliminating 99.999966% of defects from a process. If you do the math, that equates to about 3.4 defects per 1,000,000 events. Off the top of your head, can you think of any process that you control in your personal life that can conform to this level of perfection?
Ya, me neither…
Now, imagine if you were to add a few variables to the mix: multiple people, a factory, and multiple data management systems.
Then, add in a few layers of management.
What do you think your chances are of controlling a system so well are with so many variables that cannot be controlled or predicted? About 0% of the time would be a fair guess.
For a quick case study, take a minute to think about your daily work commute route and timing. Let’s say that your commute usually takes you around 30 mins. Somedays, there is bad weather or traffic, which adds a few extra minutes to your trip, and somedays fortune smiles upon you, and you fly the whole way there with record speed. For most of us, it doesn’t matter if we arrive at 8:20 or 8:40. We accept that there are factors that cannot be controlled, and we live with them. If we were to study our commute to find ways to change our own and our environment’s behavior, it would make us insane. Can you imagine the amount of in-depth planning and analytics required so that your trip would result in you getting to work between 8:29:30 and 8:30:30 as a new goal? Crazy, right?
How about another example?
Do you remember the Challenger disaster in 1986? Immediately following the destruction of the astronauts’ spaceship and death, an investigative panel was launched to determine what happened. The panel discovered that the doomed flight was given the final ok by NASA because the data available to those arguing for not launching was deemed inconclusive. The investigation found that while several engineers had only a few incomplete yet compelling data points to support their argument that the flight should be delayed, NASA decided to launch anyway. The dissenting engineers were silenced because the NASA culture at the time was characterized by a slogan in their HQ, “In God we trust, all others bring data.” Without a comprehensive data set, the engineers’ professional opinions meant nothing to NASA, and the disaster was set into motion. The expert opinions and conclusions supported by qualitative analysis meant nothing.
Thus, Six Sigma’s first fundamental flaw was that it placed quantitative analysis at a premium- the gold standard. Yet, we forget that data is not perfect; it is only a by-product of the people who control the processes, feeding the systems and generating the data. Any significant process improvement initiative’s focus starts at the very bottom, the activity(s), and the people who generate the data.
There’s a great saying that fits here that goes- “the tail doesn’t wag the dog”- data doesn’t control the business. Data is ugly, and it often lies. Suspicious or outlying data can be deleted. Keep in mind that any statistical analysis begins with cleaning the data to remove anomalies and unexplainable trends. Data is only the final output of a series of human-designed and controlled processes. It’s easy to manipulate numbers, so they say what you want them to say. Six Sigma requires problems and solutions where that fit a clean statistical model. Too much data or too little, and the problem may be overlooked or ignored.
Data + Statistical Analysis <> Problem Solving
Educated Elite
The second problem with Six Sigma was that it created a system that trained only certain people to improve process improvements. If you weren’t a green belt or black belt, you were a second-class citizen to the educated elite.
The training program, which involved advancing through a hierarchy of belt levels, would take years to master. Organizations would deploy this group of specialist practitioners from the headquarters to solve problems across the company. They would often show up at a manufacturing location, throw a solution up, declare victory, and then walk away and hope that any degree of improvement implemented would be maintained by people who had zero ownership or interest in it. As we learn in the next chapter, the ownership mentality is critical to make improvements stick.
Where’s the Beef?
The third flaw with Six Sigma was its very hierarchical and top-down-focused nature. We’ve just spent an entire chapter explaining the problems with this structure; it’s utterly contradictory to developing ownership and excellence in execution.
Before a single process was ever studied or improved with Six Sigma, an organization had to hire at least one master black belt and several black belts. Further training was then required to train additional lower belt classes-training rather than doing just consumed more resources, time, and money.
At a conservative estimate, hundreds of thousands of dollars would be required just to put a program in place. A well-accepted principle of any process improvement program or position is that the improvement should be, at the very minimum, able to cover their salary through improvements. A 2–3x ratio isn’t uncommon. Six Sigma focused on adding costs first and adding value much later.
If you’ve ever seen the classic Wendy’s commercial from the 1980s, you’ll remember an old lady driving around to different fast-food restaurants exclaiming, “Where’s the Beef?” She was tired of getting fooled with sandwiches that looked larger in advertisements but were mostly bun in person. In a nutshell, that’s Six Sigma- a big advertising budget, big corporations behind it, and only a sliver of meat sold as an overblown promise. (link below).
https://www.youtube.com/watch?v=Ug75diEyiA0
The people who work on process improvements must be the process owners- this is the basis of the Little Whats that facilitate achievement of the Big What. Not just some people, but all people must be focused on continuous improvement and operational excellence.
My Experience
I earned my Six Sigma Greenbelt while at DuPont, and it was a condition for getting promoted to Sr. Accountant. The process that I investigated involved understanding the root cause of system errors. It turns out the people at the plants were miskeying information in the required format for cost center, plant, and GL. I researched, tested, and implemented the solution. Then I approached management and asked to do the same thing for all other plants at the service center.
To be honest, my data and results never met 99.9996% of errors, so I removed some data until it did. I’m sure this happens all the time. My solution fixed the problem, and I jumped through the hoops to check an arbitrary corporate box.
What turned me off from Six Sigma is the pure amount of bureaucracy of the thing. The amount of paperwork, meetings, and box-checking stops real solutions from being investigated and implemented.
A better solution is to focus on process improvement frameworks that include everyone and focus on common-sense principles like the Pareto which 20% of drivers are responsible for 80% of the outcomes. You shouldn’t need an advanced degree or tedious hoops to jump through to solve problems and ensure that they stick.
Ben
Originally published at https://www.linkedin.com.
