The value of Process Mining

Most real-life processes have not been optimized or are not even explicitly designed. In such situations, the application of Process Mining techniques can be used to discover and unveil hidden knowledge and to improve business processes. Process Mining (PM) is a discipline that can be seen as the intersection of Data Science and Process Science and it is useful to support process improvements, understanding process performances and support optimization activities. Process Mining extracts process models from the execution logs (also called event logs) coming from company’s databases, information systems or business management software such as Enterprise Resource Planning (ERP), Customer Relationship Management (CRM), Supply Chain Management (SCM), etc.

In this paper a brief introduction on Process Mining will be given, moving the focus on why it is important to introduce it in the digital transformation of organizations. The main concepts of the Process Mining and the PM techniques to extract value from business processes will also be illustrated. The potential benefits of Process Mining can be found in businesses of any size and in any industry for process analysis and optimization, conformance validation, detecting and diagnosing bottlenecks and deviations and for process automation support. Moreover, Process Mining can be used to create AI/ML problems to predict outcome or to trigger corrective actions.

Introduction to Process Mining

How much organizations know about their operational processes? Are they able to identify bottlenecks and deviations or to anticipate problems and trigger corrective actions? Is it possible to optimize actual processes and diagnose performance and compliance problems?

Process Mining can answer to these questions. But what is Process Mining? Process Mining is one of the most innovative and exciting technique supporting organizations to provide insights from their processes allowing to support decision-making. Process Mining can also be seen as the intersection between Data Science and Process Science as it embraces the main concepts from both fields to support Business Process Management with the consolidated and cutting-edge technologies.

When theoretical work on Process Mining starts at the end of the 1990-ties, the focus was on workflow automation. Advancements in the digital universe improved data gathering and analytics technologies make it possible to record and analyze events from any sources and in any businesses. This allowed to have event data readily available and to fine-tune PM techniques. With this constant technological evolution the principles of Process Mining can be applied not only in administrative or financial fields where the workflow management systems lie, but also in production, logistics, healthcare, government, and so on.

Although the first tool for Process Mining was created in the early 2000s, nowadays there are over 40 commercial PM tools. Process Mining is used by thousands of organizations and can be applied in any domain (finance and insurance, energy, logistics and transport, healthcare and so on).

At the beginning of 2023 Gartner published the first Magic Quadrant for Process Mining, demonstrating that companies tend to use Process Mining to analyze their processes.

Why should processes be analyzed?

If you could have a tool to view and understand any business processes in your company from single activity to an aggregated view, many insights can be extracted allowing you to enhance actual process, optimize workflow, automate repetitive activities and increase the efficiency of the process.

With increasing of data processing technologies, it is possibile to access to large amounts of event logs. Every single process activity can be analyzed providing a comprehensive overview of a process. Process Mining tools can visualize process models in a detailed view of a process as well as in an aggregate perspective. This provides to evaluate the actual process based on a different insight that can be extracted.

Once the Process Mining results are approved, most of the time of the top management is spent on discussing on why things are happening rather than what is happening. To understand processes and improve them it is important to firstly analyze the reasons for delays and loops. Based on the obtained insights, actions can be defined, deployed on the actual process and continuously monitored.

Below several examples of application domains in which Process Mining can be applied to face complexity, provide new insights and support digital transformation:

• CRM: the main reason to apply Process Mining in this field is to process orders quickly and efficiently identifying delayed orders and reasons for delays.
• Manufacturing: bottlenecks can be identified in processes to optimize value stream and to reduce material waste, costs and unnecessary activities.
• Finance and Accounting: Process Mining can help to identify deviations between standard payment and confirmations.

How can Process Mining analyze and improve processes?

Before starting any Process Mining project, it is necessary to consider three complementary essential factors:

• Purpose: what is the Purpose for the usage of Process Mining?
• People: do the people involved have a mindset for change?
• Process traces: Are event logs available and reliable?

Probably the biggest challenge for every Process Mining project is related to the data availability. Data is hidden in the source systems and usually stored in a distributed way. Furthermore, data can contain confidential information and it is important to make data anonymous and not ascribable to natural persons. Data must be complete with any information that can allow a 360-degree view of the business process and the process traces must be continuously available with minimum latency for analysis and for real-rime intervention on actual process.

Based on process traces stored in event logs, Process Mining allows to analyze business processes of any domains. An event log is a set of events each of which refers to a specific activity that took place in order to perform a business process and can be assigned to a unique case. An event log must have three mandatory attributes:

• Case ID, a numeric identifier for a specific process instance
• Activity, that specifies which activity has taken place
• Timestamp, indicating the time when an activity has been taken

Further attributes can be added to provide additional information about the activities such as resource, location, cost, etc.

How does Process Mining work

Event data are assumed to be of a high-quality meaning that data have been extracted, selected, filtered and cleaned in a correct manner. Data extraction is an integral part of any Process Mining project, and probably may be time-consuming as data are stored over many database tables and they need to be converted into a specific format that can be used as input to apply PM techniques.

The main Process Mining techniques are the following:

• Process Discovery: starting from an event log, process discovery techniques automatically extract process models describing all behavior observed or just the dominant behavior. Inspecting the discovered model, noteworthy insights can be mined.
• Conformance Checking: relates events in the event log to activities in the process model allowing an in-depth analysis regarding discrepancies or deviations between To-Be processes and As-Is process flows. Deviations may be related to undesired or unusual behavior depending if the process model in normative or discovered.
• Process Enhancement and monitoring: a type of Process Mining to achieve a specific performance level and to fulfill requirements based on laws and regulations. These can be done through process extension and process improvement that allow to obtain process models with higher level of precision and without portion of models that do not meet specifications.

With advancements of technologies and with new demands in the use of Process Mining, other advanced PM techniques have emerged, such as the following:

• Predictive Process Mining: a branch of Process Mining designed to predict future activities or the completion time of an ongoing process instance. Predictive PM allows organizations to prevent undesired outcomes, delays or the violation of a regulation before they occur.
• Streaming Process Mining: processing a stream of data aims to extract relevant insights about the ongoing processes. Streaming PM is useful in fields that require a timely understanding of the behavior and usage of a system rather than improving future process instances or inspecting past ones.
• Object-Centric Process Mining (OCPM): this most recent development in the field of Process Mining allows to analyze the behavior of individual objects (such as customer or product) and interactions between them within the process. OCPM does not follow the case notation, the object-centric event data allow events to point to any number of objects rather than a single case.
• Action-Oriented Process Mining: it connects the knowledge extracted from event data to actions and it helps to respond when compliance problems or bottlenecks emerge and focusing on the improvement of the actions triggered by traditional Process Mining techniques.

Benefits of Process Mining

Process Mining techniques can be backward-looking or forward-looking and aim respectively to find the root causes of a bottleneck in a process or to predict the remaining processing time or the remaining activities of a running process.

Process Mining provides noteworthy insights to organizations about their actual processes and allows them to diagnose problems and automatically trigger corrective actions. Process Mining can also reveal performance and compliance problems improving processes by reducing costs and removing delays.

The value of Process Mining has been documented in several use cases, some of which are the following:

• CRM: analyzing event logs from orders up to payments it is possible to visualize each step of the process. Starting from these, considerable insights can be mined supporting the reduction of manual activities identifying potential automated activities or redundant activities. Thus, costs can be reduced and efficiency improved.
• Manufacturing: collecting and visualize process traces allows to perform a value stream analysis, including how and in what sequence activities are executed and how it takes to execute them. In this way bottlenecks can be found and process execution time can be improved. More information is present in the event logs and more optimizations can be conduct on the process such as reduction in stock or of wasted material.
• Finance and Accounting: the visualization of time sequence of each payment can identify deviations from standard payment terms or deviations between payment terms and confirmations. Moreover, predictions on upcoming or overdue payments can trigger actions proactively to alert from late or due payments.

Using Process Mining value may be guaranteed in the form of visualization of the model learned from data, predictions, automated decisions or any type of data visualization and statistics.

Conclusion

Process Mining is capable of providing an overview of all events and processes in an organization. Extracting event logs, cleaning them and make them in the correct format is one of the most challenge in this field. The main Process Mining techniques can provide meaningful insights using event logs and even the model learned from data. Based on these insights, weak points can be identified and decisions taken in order to optimize actual process reducing costs and improving process effectiveness. Other advanced techniques can be applied on the event logs to perform Process Mining in a predictive or near real-time way or under perspectives other than workflows. Important values can be found in any domain of any business identifying bottlenecks and deviations, support compliance, diagnose performance and support the automation of repetitive activities.