What Process Mining Is, and Why Companies Should Do It
Original article by Thomas H. Davenport and Andrew Spanyi at HBR.com
There have long been a few fundamental challenges associated with business process management, at least for as long as the two of us have been involved with the field (forty years or so, for better or worse). Two of the most troublesome problems, in our view, are at least partially responsible for the fact that process management and improvement is, among many companies, a back-burner issue at the moment. But a relatively new and innovative technology, process mining, has the capability to solve both of the problems and to revitalize process management in firms where it has lain fallow for years.
One problem involves the creation of “current state” processes — a description of how a business process is being performed today. In business process reengineering, organizations are primarily interested in an improved “to be” process, so often they have little interest in exploring “as is,” or how the process is currently performed. But understanding the current process is critical to knowing whether it is worth investing in improvements, where performance problems exist, and how much variation there is in the process across the organization. As a result, some companies tend to either skip current process analysis altogether, adopt shortcuts to it, or pay consultants a lot of money to analyze the “as is” process.
Companies that adopt an incremental improvement approach, on the other hand, tend to spend too much time on analyzing the “as-is.” In addition, their current process analysis is frequently based on interviews and sticky notes, which executives sometimes regard as overly subjective and treat with justifiable skepticism.
The other general problem with process management is the lack of connections between business processes and an organization’s enterprise information systems. Some enterprise systems (SAP, for example) are process-oriented in the sense that they support processes like order-to-cash or procure-to-pay, but there is rarely an easy way to understand how the process is being executed from the information system. Some different technologies (such as Microsoft’s Visio or Software AG’s Aris) support aspects of process design. But if you want information about how your process is performing day to day, that has typically required a difficult set of manual steps to gather and synthesize data. And many process improvement approaches — Lean and Six Sigma, for example, have not emphasized information technologies as enablers of processes or of process management.
Enter Process Mining Process mining can address both of these issues. For many years process mining has been an academic topic — ardently pursued by researchers like Wil van der Alst, a Dutch computer scientist. But the approach had little practical relevance until 2011, when Celonis, a Munich-based company, was founded. Van der Alst is the Chief Scientific Advisor at Celonis, and the company has developed four major versions of its software. It has strong ties to SAP, which is a reseller of Celonis. It collects data from that vendor’s systems, but also from others like Oracle, Salesforce, and ServiceNow, as well as any other type of system through APIs.
All of a sudden process mining is receiving increasing attention. For example, Gartner published a market guide for process mining in 2018 that included several common use cases for process mining and an analysis of the vendor community. Gartner identified over a dozen process mining vendors, with most based in Europe. Celonis was judged to be the market leader. Fluxicon (based in the Netherlands) was considered to be the most popular stand-alone analysis focused tool, and Finland-based QPR Softrware one of the oldest and more comprehensive tool sets in this space. The fact that all these firms are based in Europe suggests that process mining is somewhat further along there, although we have spoken with several U.S. companies that are employing it as well.
Process mining software can help organizations easily capture information from enterprise transaction systems and provides detailed — and data-driven — information about how key processes are performing. It creates event logs as work is done: an order is received, a product is delivered, a payment is made. The logs make visible how computer-mediated work is really happening, including who did it, how long it takes, and how it departs from the average. Process analytics create key performance indicators for the process, which enables a company to focus on the priority steps to improve. AI algorithms can detect the root causes of variation—for example, they might point out that every time a new customer needs a credit check, the process is slowed down considerably.
The selection of where to apply process mining is important. Organizations will get the best value from applying it to processes that have been digitized (i.e. supported by an IT system) and where there is still some unstructured work (i.e. reviews and approvals) that happen outside the IT system. Process improvement professionals appreciate that there’s much more than just data involved in improving complex, cross functional business processes. As Jaakko Riihinen, Senior Vice President of Products and Technologies at QPR Software, pointed out, “Process mining as a method is dependent on case and event data in the logs, and it can accomplish only things allowed by that data.”
Nevertheless, these capabilities, as you might imagine, are catnip for anyone whose job it is to oversee, improve, or troubleshoot operational business processes. Development of current state process flows is automatic and no longer labor-intensive. Even if you’re not concerned with process management overall, or even the state of the broad process in which you work, you can find out what’s happening in your little piece of it and fix any problems. If your company uses enterprise systems to support key business processes — and that’s almost every big firm these days — you should be exploring process mining.
Process Mining at Chemours The Chemours Company is a global chemistry company that was created in 2015 when DuPont spun off its Performance Chemicals segment, which includes its titanium technologies, fluoroproducts, and chemical solutions. Chemours is now an independent $6 billion company with 7000 employees, 26 manufacturing sites, and approximately 4000 customers in over 130 countries.
Chemours inherited its business processes, a dated legacy ERP system, and a keen attention to process management and improvement from DuPont. A corporate transformation effort to streamline how Chemours operates and create greater agility stimulated interest in process mining, the company acquired Celonis software, and the first end-to-end process targeted for process mining was the order-to-cash (O2C) process. The Chief Information Officer (CIO), Dean Meyer, championed the overall transformation and the Chief Financial Officer (CFO), Mark Newman, served as the executive sponsor for the O2C initiative. Sung Lee, Director of Business Process Transformation, oversees the process mining project. Lee emphasized that “the leaders at Chemours have been supportive from the beginning. This is essential for success.”
Prior to the process mining project, no one could really articulate how the entire O2C process was performing at Chemours, as people typically see just their part of the process. It took the process mining effort four months to uncover how the actual process was performing (not just what the ERP documentation stated). It made the entire process visible and revealed some glaring issues. Credit holds was one such issue, as process mining exposed that strategic customers were sometimes placed on credit hold needlessly to enable manual steps in the O2C process.
While the tangible benefits from applying process mining to O2C at Chemours are still a work in progress, quite a few issues have been identified and there are now over 40 projects underway to address key issues in O2C at Chemours and to realize benefits in the form of process simplification, communization, and automation. According to Sung Lee, process mining also contributed to an improvement in role clarity and greater cross functional collaboration; teams could see for the first time an integrated view of the end to end process, including deviations from the norm by business and product line. Chemours intends to deploy process mining in 2019 to the source to pay (S2P) process. Further, it has already conducted a proof of concept for robotic process automation (RPA) understanding the potential synergy in combining these two tools. The goal is to enable the workforce to focus more time on customer-facing activities and deeper business performance analytics.
Process Mining at ABB ABB is a technology firm with operations in more than 100 countries. It offers products and services in electrification, automation, robotization, and digitalization. ABB has a long history of quality and process improvement. In the 1980s, ABB was committed to total quality management (TQM). In the 1990s, ABB adopted the Rummler-Brache method of process improvement, and over the past two decades it has embraced Lean Six Sigma approaches. It has a culture that’s characterized by a customer-centered focus. Due to ABB’s size, there’s an ongoing search for new tools and approaches and a keen interest in internal benchmarking throughout its global operations.
The Quality and Operations department at ABB is responsible for the deployment of process mining at ABB, and selected Celonis as the preferred vendor. A small group coordinates the process mining effort in the head office, and as much as 80% of the process mining work is done by Quality & Operations personnel at the business unit level as part of ABB’s continuous improvement program. Process mining has been used at ABB to analyze operational processes related to SAP such as Order to Cash, Purchase to Pay, and Complaint to Resolution. One of the principal benefits of process mining is improved transparency of work flow, which reduces the time for continuous improvement efforts using the Lean Six Sigma method. Process mining also contributes to reducing non-value-add activities and eliminating manual reporting efforts.
According to Heymen Jansen, Head of Quality & Operations Process Analytics at ABB, there can be lot of variance in processes, as it’s a large value chain and the transparency delivered by process mining enables ABB to do more effective benchmarking across ABB’s various business units. Jansen noted, “There’s always some unit doing it better out there, where others can learn from.”
Daniel Helmig, Group Head of Quality and Operations at ABB, shared that many managers assumed that processes such as the approval for “procure to pay” — should be simple. But with operations in over 100 countries and various ERP systems in use, he and others were somewhat surprised at the number of deviations and exceptions revealed through process mining. The power of process mining may well be amplified in the future at ABB as it is integrated with other tools such as robotic process automation (RPA) and artificial intelligence (AI).
Process mining may not be for everyone. Large, complex organizations with a commitment to quality and a keen interest in internal benchmarking can best benefit from the transparency it creates. And if an organization isn’t oriented to process management, it probably won’t do well with process mining.
Process Mining and Other Technologies Process mining has been used effectively to analyze the current state of business process performance, identify areas of improvement, and assess the results of process improvements. This makes it an effective partner for tools like robotic process automation (RPA), as it can first identify the best places to implement “bots” and then provide the means to calculate the beneficial impact of the RPA implementation.
Process mining depicts a visually appealing and a data-based view of process performance. This will attract the interest of senior executives, who can easily see where problems and opportunities lie. It will reinforce an organization’s dedication to data-driven decision making. Some vendors have already identified the key steps in using process mining for greater success in implementing RPA. We expect many solutions in the future that involve combinations of process mining, RPA, and machine learning.