Anyone working in the field of process management or business optimization today is increasingly stumbling across the term “process mining”. Sounds like a buzzword at first. Maybe something to do with data mining? Is that something for the IT department? You take care of processes, not data, don’t you?
Well, yes and no.
Let’s get to the bottom of these (mis)understandings about “Process Mining”. This article will help you to understand what process mining is, how to use it and how to integrate it with classic business process management.
What is Process Mining?
Process Mining is first and foremost the visualization and analysis of event logs with the help of algorithms and mathematical procedures. Event logs are the protocol of IT-based processes. They list the events – individual activities in the IT system – together with their attributes. Typical attributes are the case ID, the time stamps of the start and end times, and other attributes of the event that are dependent on the system, such as the person processing the event or the location. An event log thus maps one or more business processes holistically from start to finish. Process mining methods are “Process Discovery”, “Conformance Checking” and “Model Enhancement”.
Process Discovery describes the data-based visualization of a process. The model is usually generated automatically from the available data of the event logs. The visual representation of the discovered model is often shown as a so-called Direct Follower Graph:
If required, the Direct Follower Graph can also be converted into a BPMN model, i.e. an abstract representation of the ideal process flow. The primary goal of Process Discovery is to create transparency and gain in-depth insight into the process flow.
Conformance checking compares event logs with an existing reference model (target model) of the same process. The comparison determines the correspondence between the target process and the lived process (actual process). This even makes skipped or added process steps and paths visible:
Meanwhile, Model Enhancement describes the analysis of the process model for optimization potentials. Especially high processing or idle times are relevant indicators. This allows you to identify bottlenecks or unforeseen process sequences and provides a basis for further optimization.
The objective of Model Enhancement is to optimize the process model and thus the underlying process. However, the process still needs to be implemented so that it becomes the new optimized actual process.
How does this fit in with business process management?
Traditional business process management usually follows the process management life cycle. This means that a process strategy must first be defined for the company. The strategy is based on the general corporate strategy, followed by the documentation of the currently lived processes. Process Mining, which generates the system processes from the log files, now helps with the documentation of the processes. Employee interviews and manual process modeling are a thing of the past.
A further advantage is that the generation always models the processes on the same level of detail. Differences in the levels in manual modeling and surveys often result from the fact that one employee perceives “Fill application” as one activity and another divides the activity into smaller process steps such as “Fill field 1”, “Fill field 2”, etc.
The documented processes are then optimized on the basis of the defined process strategy. Since analyses are necessary for optimization, Process Mining is also a useful support here. Process Mining tools automate a large part of the analysis steps. This increases the efficiency of the analyses and requires less time and resources.
In the next phase, the optimized processes are introduced into the company’s existing process landscape and implemented. As a result of the implementation, employees carry out the processes, enabling the Process Mining tool to record the performance of the processes. Then the evaluation of the first event logs only has to be compared with the evaluation of the new event logs. This means that you do not have to measure and calculate key figures every time you run the controlling function. The system analyzes the data as soon as event logs have been imported.
Process Management Life Cycle
Companies use the recorded performance data and key figures in the last phase of the cycle to check whether benchmarks are being adhered to or whether there are bottlenecks and thus optimization potential. In addition, conformance checking in this phase checks whether the processes continue to run according to their conformance.
It becomes clear that Process Mining can be easily integrated into the existing process management lifecycle. Process Mining simplifies and accelerates the documentation of processes, process optimization and process controlling. This leaves more time for the company’s core activities. In addition, many other functionalities for process analysis using Process Mining are conceivable for the future. For example, the prediction of deviations and bottlenecks based on legacy data, or even automated process correction with the help of Machine Learning.