In my previous posts I described my perspective of the future of manufacturing as it relates to cutting tools, production scheduling, inventory management, and quality management. I then described why each of those areas would be valuable to pursue. In this post, I’m going to discuss my perspective regarding how to start thinking about a digital transformation in manufacturing and where to begin.
GE has adopted a philosophy that we call Fastworks. Fastworks is based on the principles outlined by Eric Reis in The Lean Startup. The overarching idea is to think big, start small, and scale rapidly. This is accomplished via an iterative set of “minimum viable products,” which are simple ways to test a hypothesis and are designed to give feedback which allows someone to decide whether to “pivot” on the idea or “persevere”. We had this philosophy in mind when we set about undertaking our digital transformation inside of our factories. GE also has a long history rooted in philosophies like 6Sigma and Lean. Therefore, we wanted tools to enable our teams to identify and implement 6Sigma projects faster and we used tools like Value Stream Mapping to guide us towards the valuable portions of the process. In addition to identifying area/projects with high value, we found that culture is a critical consideration. Culture is critical because these initiatives are ultimately going to struggle or thrive based on how well they are adopted by people.
It is tempting to think about how digital tools will automate decision processes and potentially replace some of the processes that people do today. However, I’ve seen many “failed” improvements where people tried to take too many steps at one time.
In the essence of Fastworks, I propose working on a minimum viable product first, learning, and then continuing to the next. In many cases, this means augmenting a person’s decision process first, by making recommendations, and then, after receiving feedback and learning, gradually beginning to let the digital tools take action. For example, before trying to tackle automated feedback systems for process quality control where humans primarily control the process today, consider analyzing the quality data first to ensure that the process is understood well enough to implement control rules. In many cases, there is a significant amount of variability that people just deal with. However, if it is not accounted for in the control system, then the system will not work. Therefore, I propose the following steps:
- Analyze the current data,
- Make recommendations based on today’s understanding of the data and process,
- Analyze the data after making recommendations to observe behavior and outcomes,
- Utilize those observations to improve the recommendations over time until they are relied upon and the product outcome is always good, then
- Automate the process feedback.
Operations personnel are inevitably already working on improvement projects within the manufacturing process. They have access to some digital data, but in many cases, they spend a significant amount of time accessing multiple systems and aggregating the data or manually collecting the data that is missing. You will be able to achieve many quick wins by just enabling operations personnel with automated data collection for the data that they are manually collecting or providing software tools that combine data from multiple existing sources.
It is always a good idea to try to maximize the value that you will receive as you roll out an initiative. In the case of most improvement initiatives, I propose utilizing a value stream map to determine where constraints are and where significant time and resources are within the process. The value stream map should provide end-to-end visibility of the process for creating product. This visibility is actually more rare than you might assume.
With this visibility, you can start to assess whether process steps are “value-add” or not. If not, you can ask if those process steps can be eliminated or reduced through the usage of data and digital tools. A very detailed value stream map will show “hidden factory,” or operations that are not a part of the “official” process. You may find that the product is being measured as it leaves one process and then measured again when it enters the next process. This redundant work could be avoided by sharing dimensional data from upstream processes. This is a simple example, but one that could be avoided just by making sure that data is available everywhere that it is needed.
You can also begin to assess where constraints/bottlenecks are within the process. By focusing on the constraints in the process, you will be able to identify where to focus to increase throughput, reduce costs, or reduce inventory. By having the real situations which need improvement within your manufacturing process, you can begin to estimate the value of improving the current state in those situations and you can rank improvement projects based on their value to the business.
Value is not the only criteria to look for when deciding where to start. Culture is a critical when starting on a digital transformation. I read a post on Harvard Business Review a while back that stuck with me: Who’s Afraid of Data-Driven Management?. It describes people’s actual performance relative to their perceived performance and breaks them into groups to show how willing they would be to adopt data-driven management. In the post, the authors suggest that people who are high performers and are perceived as high performers may actually be skeptical of data’s ability to help them whereas people who are high performers but are perceived as low performers may welcome data as a way to get the recognition that they deserve. This article stuck with me because it was relatively accurate based on my experience. However, I would also add another dimension. The other dimension relates to a person’s ambition to be a top performer. If someone were in a position that had historically performed low, but they were seeking to improve their performance, they may look to data to give them an edge for making improvements.
Some people may doubt the value of data, some may be afraid of what the data will show, and some may welcome it with open arms because it will give them validation or the ability to identify and drive improvements. When deciding where to start a digital transformation, you should consider the types of people that you will be engaging.
If you encounter a skeptic, they may only support the initiatives half-heartedly. If you encounter someone who is afraid of the data, they may place roadblocks in front of you or attempt to sabotage the initiative. Ideally you would seek out and engage those people who will be advocates and help first. Having initial successes with them will intrigue the skeptics and allow you to be better prepared to work with the people who are afraid of the data.
In this post I’ve discussed my perspective for getting started on a digital transformation within manufacturing. I’ve given some general considerations, discussed my thoughts for finding value, and provided some considerations for culture. In my next post, I will discuss machine monitoring as a starting point. Please come back to read that post.