The successful application of process mining relies on good tool support and data availability. However, this is not enough. To illustrate this, I review the so-called L* lifecycle model for process mining.
CRISP-DM, stands for the CRoss-Industry Standard Process for Data Mining. This is a methodology for conducting data mining projects. CRIP-DM identifies a lifecycle consisting of six phases: (a) business understanding, (b) data understanding, (c) data preparation, (d) modeling, (e) evaluation, and (f) deployment. These phases are very general and provide little support for process mining projects. Therefore, the L* lifecycle model for process mining was developed and first described in my first process mining book, which appeared in 2011.
The L* lifecycle model has five stages, starting with stage 0 and ending with stage 4.
Stage 0 is called "Plan and Justify." Like any project, a process mining project needs to be planned carefully. For instance, activities need to be scheduled before starting the project, resources need to be allocated, milestones need to be defined, and progress needs to be monitored continuously.
Stage 1 is called "Extraction." After initiating the project, event data, models, objectives, and questions need to be extracted from systems, domain experts, and management. We already discussed the challenges of data extraction. However, this stage also aims to extract additional artifacts and knowledge, for example, definitions of Key Performance Indicators and hand-made process models.
Stage 2 is called "Create Control-Flow Model and Connect Event Log." The control-flow, that is the ordering of activities, forms the backbone of any process model. Therefore, first, the control-flow model is discovered and verified. After completing Stage 2 there is a control-flow model tightly connected to the event log, i.e., events in the event log refer to activities in the model. The output of Stage 2 may be used to answer questions, take actions, or to move to Stage 3.
Stage 3 is called "Create Integrated Process Model." In this stage, the control-flow model is enhanced by adding additional perspectives, for example, the organizational perspective, the case perspective, and the time perspective. The result is an integrated process model that can be used for various purposes. For example, the model can be inspected directly to better understand the as-is process or to identify bottlenecks.
Stage 4 is called "Operational Support." This stage is concerned with operational support activities, for example detecting deviations of running cases, predicting delays, and recommending countermeasures to avoid predicted performance or compliance problems. Operational support is the most ambitious form of process mining and can only be achieved if the process is structured and rather stable. If predictions are unreliable, one should not use them. Moreover, process interventions sometimes have an adverse effect. Therefore, it is important to continuously monitor the process. Fortunately, this is possible with today's process mining software.
The L* lifecycle model describes the five stages. It is important to have the support of higher-level management. Bottom-up initiatives are good to build up experiences and provide proofs-of-concept. However, process mining is most effective when it is done continuously for many processes. The business case for a pilot project is problematic because, typically, 80% of the initial efforts are used for data extraction. The most significant return on investment is achieved when process mining is a continuous, large-scale initiative.
Another reason to seek the support of higher-level management is that process mining often reveals inconvenient truths. Full transparency may reveal severe compliance problems and unexpected bottlenecks. Therefore, bottom-up initiatives may be successful in showing possible improvements and still get blocked by middle management. The bigger the improvement potential, the more painful process mining diagnostics may be.