Don’t let chaos dominate your work life. If work feels chaotic or an urgent problem is rapidly consuming resources, it is time to step back and assess how teams are working. Work ideally should look like a world-class orchestra performing. Talented, coordinated, synchronized, with a rhythm. Often, it looks more like a disturbed beehive. Fast, urgent, chaotic, and everybody running around like they are late for something very important. Sometimes, simply observing how people are working reveals an underlying problem.
I was asked to help a team that was dealing with a disturbed beehive moment. One of our new technology silicon devices was yielding less than expected, which sent teams scrambling to develop new plans based on this information. Yield is calculated many times as silicon wafers are processed and eventually put into a final product like an SSD or Solid State Drive. There is line yield in the factory, then wafer yield, package yield, and, eventually, end-product yield at the SSD level. In manufacturing, there is always some fallout or yield loss along the way; nothing can be perfectly made. At a high level, yield is the ratio of good vs bad. A 50% yield means that half are good and half are not. As silicon technologies are developed early in their life, yield starts low and then increases over time as teams learn how to manufacture a new product on a new process technology. There is usually a plot projecting yield over time that teams use to develop their plans as the technology ramps over time. These plots are also used to allocate material for continued process and yield development, test development, design validation, and eventually customer samples.
We had a spreadsheet model for NPI (New Product Introduction) material that analyzed supply, demand, and yield to determine whether we had enough material to supply all our teams and eventually customers. In this case, there was not enough material for everyone. So, who got shorted and how much was the question of the day. Keep in mind that customer ramp plans and the company's revenue projections depend on programs being delivered on time and on budget. To make all of that happen, teams were turning over every rock, looking for ways to get more material into the pipeline, and adjusting their demand to align with projected supply. Everybody was adjusting their plans real time, and new information was coming from the factory (how much can be produced), engineering (how many were needed for testing and validation), and customers (how many samples they needed to develop their systems). The supply chain team was running their model, 2 to 3 times a day to improve their forecast accuracy, so teams could use the latest and greatest data. The model output then would determine if supply met demand or if demand was greater than supply. It was a beehive moment, and there was not enough supply for the current demand.
The teams were overloaded from working on this problem, and the work process they were using was not effective; it had them all scrambling. We needed to modify the current workflow into a more organized and coordinated effort. It didn’t have to be like this. To help, we first needed a baseline or current state analysis of the work process to see exactly what was happening and assess what we could do to make things better. We developed a current state workflow map to help us visualize what was happening. After that, we looked at the main breaking points or where some challenges were in our workflows. Then we started building a new future-state workflow to meet the moment.
When we analyzed the current state map, the first discovery was that 2 to 3 supply, demand, and reconciliation model runs per day were excessive. Second, the request for model run updates from all stakeholders was not coordinated, resulting in numerous daily updates across multiple meetings and for multiple leaders. Figure 1 below is the current state high level workflow. The yield ramp would take months, so we needed a work process that would endure. There was no quick fix available, and there usually isn’t, since work today is never that simple.
Figure 1
Based on the challenges the team was seeing, I helped design their future-state workflow with the team leads. The highest priority was to reduce the number of supply, demand, and reconciliation model runs that the teams were producing and consuming. This is an asynchronous workflow problem where work across teams is not synchronized. This was causing an unbelievable amount of work for the teams. I proposed we establish a weekly cadence; I thought that frequency was sufficient for planning purposes, and we didn't need intraday updates. We would run the model once a week and coordinate the inputs so that all teams could update the supply and demand data simultaneously. Here was our initial design. By Monday at noon, all teams were to deliver updated supply and demand numbers. On Tuesday, the model team would review the inputs and clarify anything needed to ensure the model run would be accurate. On Wednesday morning, a new model run was completed, and the data was disseminated to the teams. On Thursday morning, the team would decide whether an escalation was required based on the new model run.
Escalations were another challenge. We discovered there were three distinct escalation types, not just one; we needed three different leaders, one for each type. This is a typical problem in workflow design: a single process is used for many different things. This leads to workarounds, as people have to circumvent the current workflow design to do their jobs. As we worked through this, we helped the leaders align on what would be needed for each type. The leaders were not on the same page either because they didn’t have the details needed for the different decisions that had to be made based on the model's data analysis. The 3 escalation paths were: Customer impact (fewer samples – customer risk), supply chain impact (NPI versus Production trade-offs or wafer allocation), and development team impact (less testing or analysis – product quality). For all these potential paths, we integrated our process into existing meetings. Then we tested the process, educated teams and leaders on it, and made a few adjustments as we learned. Figure 2 below shows the new workflow, or future-state design. Each time we ran the new work process, we discovered tweaks we needed to make, then tested it again. After a few iterations, the bees were back in their hive and back to their daily routines; the scramble was over. We had a new work process in place to address the situation at hand. In a few weeks, I moved on to the next workflow challenge.
Eventually, yields rose, production ramped up, and we delivered samples to customers. In product development, beehives emerge every day somewhere, and we deal with them the same way. Draft a problem statement and expected outcome. Map out the current state work process. Evaluate the current process for where it is breaking down. Then designing the future-state process and testing it. Then tweaking it until it was good enough. Each time a work process is used, it tests the current work process. If done right, a documented work process is a hypothesis, and its execution is an experiment. Integrating the scientific method into how we work is key to managing change and complexity in today’s workplace. This makes each execution run a learning cycle.
We turned a beehive into an orchestra. We developed a workflow to meet the moment, and one that could be used for the next technology ramp. This helped teams get their work done, reduced the need to work nights or weekends, and relieved stress for the people involved. We aligned leaders on the challenges and the fixes. We improved data accuracy and coordinated communication across global teams. In complex product development and engineering workflows, beehives are just part of it. Something is likely to stress-test your workflows, and people should flag it when they see it. Hopefully, after the beehive is contained and everything returns to normal, there is a little honey left over for everyone to celebrate the transition from chaos to an orchestral-performing workflow.
Do something today to improve your work-life balance. You won’t regret it. Have a great day, and good luck on your work-life journey.