“There is much more to process mining than meets the eye”

Process mining has become trending, with companies investing heavily in hiring experts and purchasing costly tools to discover, analyse and optimize their work processes. However, Vinicius Stein Dani, a researcher at Utrecht University, has found that while there is plenty of evidence that process mining can effectively help organisations, this is not always the case. His research identifies key challenges and specific causes why process mining initiatives fail, offering recommendations to overcome them and maximize effectiveness. Stein Dani defended his thesis on 21 November.

Process mining is a powerful technique for analyzing and improving work processes within an organization. By leveraging event logs -a specific data format for process mining- it bridges the gap between traditional business process management and advanced data analytics. In recent years, process mining has surged in popularity, driven by the growing demand for automation, cost savings, and process optimization. Its continued growth is fueled by organizations' need to enhance efficiency, streamline operations, and remain competitive.

The dark side of process mining

When Stein Dani began his PhD research, he initially focused on the data preparation phase of process mining. However, the COVID-19 pandemic disrupted his plans, making it impossible to visit the companies required for his project. Forced to change direction, he turned first to a literature review. During this study, he noticed an overwhelmingly positive portrayal of process mining in most case studies. This raised his suspicions. “It was just too good to be true,” Stein Dani reflects. As a result, he started interviewing process mining experts from around the world about their experiences. “That’s when I started uncovering the dark side of process mining”, he says.

No magic wand

During the interviews, many experts shared stories about process mining failure in organisations. They experienced these cases first-handedly. Stein Dani gathered the accounts and started identifying the main challenges and specific causes of these failures, which he calls ‘terminators’. 

“The main finding is that there is much more to process mining than meets the eye”, Stein Dani summarizes. “Companies sell process mining tools as if they were a magic wand”, he says, “but there is much more work involved before and after process mining techniques can be used, which people often underestimate.”

Companies can deny the findings, particularly when they are not prepared to accept that they were doing something wrong

Terminators

There are five main causes why process mining insights are not translated into actual process improvement, which hamper process mining initiatives, ultimately leading to its termination: laborious data preparation, loss of interest, lack of expertise, lack of incentive, and denial. The latter cause refers to the denial of the analyses’ outcome. “Companies can deny the findings, particularly when they are not prepared to accept that they were doing something wrong. In some instances, they even hid the results”, Stein Dani elaborates. “Indeed, process mining insights can be confronting.”

Text continues below image

Managing expectations

The identification of challenges and terminators led to a set of recommendations for companies and the process mining industry. Challenges include the establishment of organisational commitment, and availability of expertise on process mining and change management. Furthermore, the management of expectations plays a major role in the outcome of process mining projects. “Companies believe that process mining will find the bottlenecks and solve them”, Stein Dani says. “Insights can indeed lead to the improvement of work processes, but the company itself needs to implement them.” 

Further work is needed to develop strategies to deal with these challenges and the terminators, Stein Dani says. For example, one idea is the employment of large language models to facilitate streamlining the data preparation phase. This AI-technique could potentially make this phase less time consuming.