Is Your Business Filled With Kryptonite?
“Sometimes” is to a factory or process what Kryptonite is to Superman. If you’re not a comic book fan, Kryptonite neutralizes Superman and renders his powers useless. Think about the last time you started a new job. There was a series of steps that transformed you from an outsider into a new employee. Depending on the size and maturity of the company, it may have been a highly prescribed and well-orchestrated process, or it may have been an unceremonious series of events that didn’t seem like a process at all.
The latter was my experience a few years ago. I arrived on day one and the receptionist handed me an RFID badge, a laptop, and a sticky note with the WiFi password on it. I logged in, downloaded Microsoft Office, and asked around until I found someone in the company with admin rights to activate my account. A few weeks later, another team member joined. She was given a laptop with the WiFi already configured and software already installed. The following week another new team member arrived. Unfortunately, the person who ordered the computers was not informed, so this person had to wait a few days to get set up.
This continued for a few more cycles before my team of continuous improvement experts intervened. Perceiving the root cause as a simple communication problem, they injected a little technology and began digitizing a checklist. They set up automated email triggers, so the right people knew the right action to take at the right time—easy-peasy. There was only one problem: sometimes.
Processes need to be stabilized first if an improvement effort is going to be effective and sustainable.
Sometimes human resources notified the receptionist of a new hire. Sometimes the hiring manager handled everything himself and didn’t tell anyone, so a new hire would just show up. New hires usually started exclusively on Mondays, but sometimes they started on a different day. The receptionist was the one that usually created the account credentials, because she was one of the only people that knew how to do it. Sometimes she was busy with other duties when the new person arrived. The checklist my team created quickly turned into a complex logic tree. The email messages that needed to be sent had so many possible variations that we had to start numbering them.
Stability is a prerequisite for effective and economical process improvement. Imagine your processes are represented by porcelain mugs. If improvement is a liquid you can simply pour in, then just like water, the improvement will follow the path of least resistance. If you want it to remain in the mug, you have to ensure there are no cracks. Cracks, even the smallest ones, represent variation in the process, and your improvement will find a way to leak out. It may happen slowly and initially be undetected, but eventually, you will have to pour more improvement into the mug.
In 2013, The Boeing Company decided to automate the arm-vibrating, shoulder-breaking work of riveting the 777X airplane together. Riveting is noisy, usually requires two people, and takes a toll on the human body, so it made sense to transform the process by handing it over to automated machines. Boeing invested in a small army of robots that would, if successful, autonomously move around a fuselage, drill the required holes, insert the correct fasteners, and rivet them into place with fewer injuries, higher productivity, and better quality. There was only one problem: sometimes.
Six years into Boeing’s 777X transformation, the robots got fired. According to the Seattle Times, robotic automation “never delivered its promise of reduced hand labor.” Why? Sometimes the fuselage pieces were not exactly the same as the previous ones. Sometimes the holes drilled didn’t align or were oversized. Sometimes the drill bit broke. Sometimes the fastener dropped. Sometimes the vision system on the robot couldn’t find the right place to drill. What seemed like a mundane and highly repeatable task, in fact, contained a lot of variation. The metaphorical mug for the fuselage assembly process was riddled with small, undetected cracks.
Variation in a production system—whether that system is digital or physical—often goes undetected because the people performing the tasks adapt to it. The receptionist isn’t told about a new hire starting on a Wednesday instead of a Monday, but she overhears the hiring manager discussing it with a colleague, so she orders a computer. The mechanic installing a rivet sees that the holes drilled are slightly larger than expected, so he selects a bigger rivet. Robots don’t naturally adapt; they have to be taught. The more combinations of possible events, the more teaching has to be done.
When events sometimes happen, or a particular person usually does something, it’s an indication of instability. Processes need to be stabilized first if an improvement effort is going to be effective and sustainable. If the improvement involves automation or an injection of technology, instability will quickly drive up the cost. Stability does not imply perfection or efficiency, only repeatability. A stable process may produce mediocre results but will produce those results reliably. Automation, whether via Kuka robots, robotic process automation (RPA), or some other technology insertion, will not fix a broken process, it will simply repeat it reliably. Injecting technology should be a way to complement a good, stable process after all the Kryptonite has been removed.
Here’s a method for finding Kryptonite in your factory or process: create a process map. Ask all the people who perform tasks for a given process what they do. Capture their answers in the form of labeled boxes and connect them together with arrows in the order they are performed. Every time someone discusses a task with the words “sometimes,” “usually,” or “if,” add another box and connect it. Mark those boxes with a “K” or draw a little picture of Kryptonite on them. Before starting an improvement effort on the process or injecting any technology solutions, make sure you understand what causes the variation and eliminate it if possible.
