Business management is one of the great challenges that companies face year after year, where business resource planning or customer management explicitly entails the need for interconnected systems regardless of the activity or sector in which that are submerged. Thus, with the aim of optimizing operating processes and the constant search for operational excellence, a new line of work has emerged: ensuring a balance between (a) introducing, updating and/or replacing systems and (b) reducing costs derived from internal processes.
Contrary to what might be thought at first, the complexity of the systems affects the appearance of execution gaps, being one of the main causes of operational inefficiency. Process mining is a technique used to analyze processes through their real behavior, obtained by the fingerprints of the source system that supports it; for example, we will be able to know the life cycle of an invoice since it is registered in SAP until it has been filed by the financial manager. Therefore, through this methodology, the data from the source systems are converted into information structured chronologically, allowing the history of each case (or invoice) that has started the process in question to be obtained.
Process mining is a technique used to analyze processes through their actual behavior
This technique emerged at the end of the 1990s in Wil van der Aalst’s research as a new field that linked data science and process science, converting event data into relevant knowledge about the process and action proposals. At that time there were limitations to the application of mining due to the lack of data availability. However, today huge amounts of data are generated daily and almost all processes leave their traces in different computer systems such as ERP, CRM, etc. This creates a great opportunity for analytical methodologies such as process mining to win their space and the new challenge focuses on the generation of relevant information, information of value for the business that allows acting on the processes themselves.
To make adequate decisions, it is increasingly essential to rely on a solid data model. Unlike other types of data science techniques, process mining incorporates into data models information that is very important for the discipline of Business Process Analysis: order. As in Markov chains, the activities carried out during the process are not independent events, but rather depend on the predecessor event. Therefore, it is necessary to incorporate such information in the models that estimate the risk of delay in the delivery of a product to the final consumer. An important difference between Process Mining vs. Business intelligence is the scope of data models. While Business intelligence tends to use a static model of the process as a starting point to calculate the target KPIs and then analyze their evolution, mining focuses on defining a model that analyzes the process from start to finish and in real time to be able to not only to analyze the historical ones but also to influence the ongoing cases.
Some process mining providers provide relevant information and have significant synergy with other emerging technologies. For example, applying process mining can facilitate automation and ensure better results through the identification of robotic process automation (RPA) opportunities. In other words, with this methodology, it is possible to objectively find and quantify which are the activities that present the greatest failures due to human intervention or that entail a greater cost and use this information to guide the automation efforts, instead of only having information subjective of an ideal process model. In addition, it can be combined with artificial intelligence algorithms to identify the root causes of failures in the process and use such information to train Machine Learning models that avoid these failures in the future.
Table 1: Celonis Company Report
A good mining tool, therefore, must be a more powerful visualization tool than any spreadsheet and, at the same time, a more user-friendly environment than any software or other environments specialized in branches of data science. . Without that simplicity, it would be very difficult to engage end users hindering adoption of the solution and implementation of changes. However, this simplicity must be given without forgetting the technological synergies and computational needs. For this reason, many of the solutions on the market specialized in process mining developed their own PQL (Celonis) or PMQL (ProcessMaker) language. The adaptation of programming languages to process mining has several objectives, mainly to achieve an intuitive language with low computational cost and that can be used and reused in several relevant calculations without requiring a high level of programming knowledge (low-code). Facilitating the continuous use of the solution by different types of users.
By Larissa Soares Dos Reis – Senior Consultant Capgemini Engineering and Lorena Suárez Toribio – Associate Capgemini Engineering
George is Digismak’s reported cum editor with 13 years of experience in Journalism