Have you ever wondered where all the Lean, Six Sigma, continuous improvement, and operational excellence practitioners have gone? When the BPM (methodology + software) conversation turned toward digital process automation (software and very little process methodology) much of the continuous improvement conversation went away. Poof–it disappeared, to be replaced by a lot of technical talk. Nothing wrong with that but where did the practitioners go? I talk with business technology leaders all the time and get a steady stream of vendor briefings but none of them ever mention continuous process improvement. Instead, the conversations take a hard turn toward RPA, AI/ML, application integration, and low code.
I recently asked an IBM exec about why no one focused on digital process automation talks about continuous improvement anymore. In his opinion, continuous improvement is getting much harder to do with micro-services, cloud computing, dynamic case management, and significantly more complex business processes. He believes that AI/ML technology is needed to perform true continuous process improvement based on process event data from operational digital process automation systems. I think that’s true.
This week I found many of the AWOL process improvement and operational excellence practitioners at PEX’S Process Mining online conference. What a lineup of speakers (!), ranging from consultant practitioners, operational excellence leaders, and sponsoring vendors. Here is my summary from the opening keynote by Prof.Dr.Ir Wil van der Aalst (RWTH Aachen University), who is considered the father of process mining:
- Processes usually follow the 80/20 rule. Professor van der Aalst opened by saying that 80% of business processes are usually structured, with 20% of the variance eating up 80% of the costs and effort. In structured processes, the goal is to drive out variance. In dynamic processes, 20% of the most time-consuming cases may be the highest value customers, so it’s a different perspective. (Think dynamic case management.) These highly variable processes are harder to tackle with process mining. The overall goal is to remove operational friction from the last (inefficient) 20%, making structured processes the natural target.
- Process mining = data + processes. Although process mining is maturing, business acceptance is still slow. Celonis is the market share leader, along with 30 others (many of them open-source.) He advises that organizations look at process mining from a continuous perspective, not a project with a fixed beginning and end. In addition, process mining must be more than a data scientist working in a corner. The process improvement practitioners must be part of the team too.
- Process mining is very generic and adaptable. It can be applied to ERP, CRM, patient flows in hospitals, warehousing, manufacturing production processes, etc. Even highly structured processes, such as order-to-cash, have thousands of variants (e.g. hand-offs to email, phone calls) that create performance and compliance problems. But start with event data from existing processes– not process modeling. Then, after you’ve done process mining, identify what processes are super deviant and begin modeling.
- Process mining is very different from AI. Specifically, a neural network is very different from a process model. If you are trying to improve a process, neural networks don’t apply because they work with supervised training models. But, after the business process is modeled, then it’s straightforward to generate a machine learning problem that identifies performance and compliance issues. Many AI tasks are very narrow and very specific. He believes that AI will not play a big role in process mining but will be relevant in the second phase—after you have addressed bottlenecks. In summary, organizations can use AI/ML in process automation, but process mining and AI/ML are very different approaches.
- Process mining + process modeling must come together. Originally, process mining was applied to historical data. Now, process modeling typically applies to data that is refreshed every hour. The goal is to predict how the processes will behave tomorrow –not easy to do especially with processes that change. Organizations also use modeling to compare a process over time or compare the same process (e.g. ERP) that executes in different parts of the organization.
- Pursue process mining from the top down. To be successful, avoid projects that start in low levels of the organization and with an unimportant process. Instead, find a business champion and tackle processes that are meaningful. Process mining projects have high ROI so find the type of structured processes that can showcase the benefits of process mining to senior management.
My conclusions: Process mining is an esoteric topic that is hard for businesspeople to grasp. I found myself tuning out when the analysis got in the weeds. But process mining is an invaluable tool in the Lean and Six Sigma practitioner’s toolkit. To get value, it’s important to not only involve data scientists, but also involve continuous process improvement practitioners and methodologies. However, this alone is insufficient. The continuous improvement practitioners need to reunite with project teams that are tackling complex business process automation projects. Otherwise, businesspeople and technologists will stay focused on process automation tools (RPA, digital process automation) and ignore the valuable insights from process mining.
Large manufacturing companies, such as Intel and Siemens, are getting jaw-dropping ROI in the millions and millions of dollars for short process mining engagements using a handful of staff. Tackling process mining in such highly structured environments (like manufacturing, pharma, oil, and gas) is a no brainer–it will have high payoffs and process excellence will truly shine. Because those companies are already engineering-focused, it will be a much easier sell to business executives than it would be to, say insurance or banking. One company gets so many millions in savings for each process that they no longer do ROI analyses. That says a lot about the value of process mining in the right environment. If your organization has high value, highly structured processes, then process mining is definitely worth investigating.
Work with us to ensure you are a disruptor not one of the disrupted!
Get trusted advice and technology insights for your business from the experts at Deep Analysis. [email protected]