Business Process Optimization
with
Intelligent Process Twins

Finding the right degree of standardization is the key to optimizing business processes.

According to Dr. Marlon Dumas, one of the biggest challenges companies face when optimizing their business processes is how to treat each of their customers uniquely, while not having one distinct pathway for each of them. Marlon is a distinguished researcher in the area of Business Process Management, co-founder of the process mining company Apromore, and co-author of the book ‘Fundamentals of Business Process Management’. “The biggest pain point for companies when optimizing their processes by far is the lack of standardization,” Marlon states. “If every customer is going to be different, there is no other option than spending a lot of time looking into every case, taking decisions, reworking them, and doing the same for the next customer, but doing it in a different manner. That is a bit of a vicious cycle that a lot of organizations get into.”

Process Standardization and Differentiation

While more standardization is obviously a solution to handling customers more efficiently, that certainly does not mean that Marlon advocates a one-size-fits-all approach. “You need to constantly keep standardization in mind as a desirable goal, but that does not mean optimizing for one happy flow. There may be many.” Marlon sees predictive and prescriptive process monitoring as useful tools to determine which cases require us to deviate from standard pathways. “Using predictive process monitoring we can determine, for example, which actions we should perform or which pathways we should follow to best handle each case.” He goes on: “Assuming that all the variants of a process can or should be optimized “by design” is unrealistic. There are too many exceptions to pretend we can even count them beforehand.” In the end, predictive and prescriptive process monitoring allow us to optimize at two levels: We optimize the most frequent pathways “to death” at design time, and then we smartly deviate at runtime based on the specific characteristics of each case.

Intelligent Process Twins

Through his company, Apromore, Marlon sees that companies increasingly look at combining process mining, simulation, and predictive analytics to provide real-time insight into what is happening in the organization and what is likely to happen in the near future. He calls this technology trifecta the “Intelligent Process Twin” because it brings together elements of Artificial Intelligence and Digital Twinning. Addressing the observation that in business process management, organizations have historically been skeptical about the usefulness of simulation, which is at the heart of digital twinning, Marlon notes that companies are gradually recognizing the value of simulation. He believes that the past skepticism may stem from the fact that building good simulation models by hand is incredibly hard. But we are now starting to master the challenge of discovering accurate business process simulation models from data. “Combining process models with process execution data is pivotal in building accurate simulation models.” Marlon underscores that over 50% of Apromore customers are using data-driven simulation, and in general, those companies that use simulation with process mining are the ones that achieve the highest business value. When it comes to doing what-if analysis of your business processes, simulation is simply the best tool. "When people are skeptical about using simulation, I ask: what is your alternative?"

Looking Towards the Future: Online Process Optimization

Marlon sees that a convergence between the fields of Operations Research and Business Process Management is already happening, particularly in offline optimization. The frontier, however, lies in online process optimization. As processes become more streamlined, the focus shifts to optimizing decisions in real-time, a space where predictive and prescriptive process monitoring will play a crucial role.