Behind this is a technology that collects and analyzes company data to provide a completely objective and continuous view of process flows based on this data. Specifically, companies expect more process transparency, process improvements, and cost savings, among other things, by leveraging automation potential. This article aims to show what exactly lies behind this technology and what is required for implementation in companies.
What exactly is behind Process Mining?
The Process Mining technology developed by Wil van der Aalst aims to identify, monitor, and of course improve real AS-IS processes by extracting information from so-called event logs – automatically generated activity logs from IT systems.
In contrast to a classic top-down approach as is common in process management (operational application systems are implemented or adapted based on a modeled process), Process Mining in conjunction with Data Mining provides a bottom-up approach. More specifically, Data Mining methods can generate process models corresponding to reality from collected data, to potentially compare them with an existing target process model. In particular, the following questions regarding a process or process landscape can be answered:
- What really happened in the past?
- Why did it happen?
- What could happen in the future?
- When and why do people deviate from the process?
The following diagram illustrates in detail the relationship between Process Mining and a company’s IT systems. The information flow here also shows once again the bottom-up approach of Process Mining.
Relationship and information flow between Process Mining and IT systems
Here, the three core functions of Process Mining are already apparent:
- Process Discovery: Discovery of AS-IS processes from large amounts of data
- Conformance Checking: Compliance check between target and AS-IS process derived from data
- Enhancement: Process improvement
As seen in the graphic, all these three core functions are based on the aforementioned event logs. These represent the pivotal point of the entire technology and are generated by application systems such as workflow management, ERP, CRM, or other inventory management systems.
The mentioned systems automatically log an enormous number of events, regardless of whether the process activity is automated or executed through manual input by an employee. Information recorded and stored in event logs includes, among other things, details about the process instance, a timestamp, and an activity designation. The logging of events can take place in different forms. Storage is possible in database tables, message logs, mail archives, transaction logs, or other data sources. To maximize the benefits of Process Mining, the quality of these event logs must be very high. But before going into more detail about data or event log quality: How can this benefit manifest concretely?
Opportunities for Companies through Process Mining
First, there is increased process transparency to mention. Through better insight into the ‘real’ world, various potentials can be discovered and realized, for example in combination with other automation technologies such as Robotic Process Automation (RPA). This improved transparency of business operations can also drive process performance and redesign of processes, and identify ‘business process waste’. Closely related to this are cost savings through faster turnaround times and evidence-based decision-making processes. However, for all these possibilities of Process Mining to be realized in the company, some prerequisites must be evaluated beforehand.
Prerequisites for Implementation in the Company
First, despite possible conflicting goals, very close cross-departmental cooperation between IT and business must be ensured to involve all necessary stakeholders. Within this cooperation, the two groups are responsible for special areas of responsibility, which are closely interlinked.
The IT department is responsible for extracting and modeling data from IT systems, as well as providing the necessary project infrastructure. At this point, it is also important to check whether the IT staff have a sufficiently good understanding of the IT systems in the company to undertake the extraction and modeling of the data. The business, on the other hand, drives the project forward with dedicated process excellence, thus enabling the desired transformation.
Experience shows that projects tend to fail when there is no corresponding alignment between IT and business areas. A global survey among Process Mining experts from practice and research in 2021 once again highlights the clear areas of responsibility of business and IT, as well as the importance of a cross-functional alignment in Process Mining projects.
Own illustration of the responsibilities of business and IT according to a survey by Deloitte
Another important point, as in many other business analytics projects, is data availability and its quality. First, it must be evaluated whether the company has sufficient data available. Once an adequate amount of data has been provided, the sufficient quality of the data must then be ensured. Due to the great importance of data quality, a number of assessment criteria have been established here:
- Reliability: Recorded events actually took place and the attribute values of the events are correct
- Completeness: No events are missing in the recording and log entries contain necessary information (at least event, timestamp, process instance)
- Semantics: Each recorded event should be unambiguously interpretable
- Security: Wherever required, data has been anonymized
- Validity: Data reflects the real process in the correct way
o Example: A customer service process where the service representative asks the customer some questions to help solve the problem. The resulting ticket is often created only after the conversation and immediately closed again. In this case, the timestamps of the event would not reflect the actual duration of the event. If one now looks at throughput times in an AS-IS process model obtained through Process Mining, for example, this can lead to misjudgments.
Based on these criteria, five maturity levels for event logs can be distinguished, which are categorized from poor (handwritten logs, post-its, etc.) to excellent (logs from workflow management systems) quality. In summary, it can be said at this point that qualitatively high-value results inevitably require qualitatively high-value input.
Another prerequisite, which in turn influences all other success factors, is appropriate leadership commitment. Strong support for Process Mining projects by management helps to avoid resistance from individuals or teams and can even gain additional supporters and advocates for the project. It thus considerably facilitates change processes. Furthermore, management support can promote cross-departmental communication and accelerate decision-making. At this point, it is already clear that Process Mining is much more about enabling the organization and having a positive impact on the business than about the technology itself.
Looking again at the resources involved in a process mining project, their availability and methodological competencies should be ensured due to strong dependencies. Process knowledge, analytical skills, data engineering, process modeling, and management skills are essential for successful project execution. To ensure this, establishing a Center of Excellence helps. This not only ensures resource availability but also promotes a stronger alignment between IT and business. This, in turn, increases the likelihood of management support and reduces risks such as project delays or even project failure.
The factors listed are only the most common and established prerequisites in practice. Of course, this collection must be expanded to include company-specific factors. These could include, for example, the definition of KPIs or the availability of digital processes.
Conclusion and Outlook
In summary, one can say that the technology offers great potential for companies – especially process transparency and associated cost savings. However, if you want to realize these potentials, certain prerequisites must be checked in advance. Crucial points here are data availability/quality and sufficient management support. Data availability, in particular, is only given in practice in most cases from a certain company size onwards. As a benchmark, there should be a company size for which the use of enterprise software (ERP, CRM systems) is worthwhile and ideally already implemented. Furthermore, it is evident in practice that the technology is currently used almost exclusively in supporting processes (accounting, order-to-cash, service management, etc.) and not in core business processes. This is due to the fact that the efficiency potential in the aforementioned supporting processes is usually significantly higher, which, however, needs to be questioned. The previously mentioned process mining survey also reinforces this fact in the following diagram on current application areas of process mining.
Current Application Areas of Process Mining
From a research perspective, there are still some gaps to be closed in the future to further advance the technology. These include, among others, automated generation of event logs (event log preprocessing) or data quality measurement to check the suitability of process mining in the respective company. AI can be a decisive driver in this regard.
