How Process Mining Compares to BI
You may have wondered what exactly the difference is between Process Mining and Business Intelligence (BI). I get this question all the time. Is Process Mining just old wine in new skins, or even about to replace the “old-fashioned” BI? Here is my take on the topic.
I find the above picture a little ugly yet informative in illustrating the different ingredients that usually play together in Business Intelligence technology.
First of all, the extraction of data plays a huge role. According to D. J. Power1,
The term BI is a popularized, umbrella term coined and promoted by Howard Dresner of the Gartner Group in 1989. It describes a set of concepts and methods to improve business decision making by using fact-based support systems.
Second, to be able to support business decisions, the data from different source systems need to be consolidated. They are usually stored in some data warehouse, from which reports can be generated or queries can be answered (e.g., using OLAP).
How does Process Mining compare?
Process mining fits in on the analytics side of the whole BI landscape (so, on the right side of the picture above). It has no particular methods to offer that help with the extraction or management of data. Since the consolidation of different data sources is also crucial for process mining in order to analyze end-to-end processes, existing BI technology could be leveraged here.
In my view, the differentiation of process mining with respect to traditional BI is twofold:
No. 1: The added value of process mining over traditional BI reporting tools lies in the depth of the analysis.
Traditional BI reporting tools focus on the display of Key Performance Indicators (KPIs) for executives in the organization. For example, the cycle times of a customer-facing process may be key in meeting certain service levels that have been agreed.
If the cycle times are out of the acceptable bounds, dashboards can highlight this problem. However, they cannot do much to uncover the root causes for this problem.
Process mining can help to provide much deeper insight into the actual processes by uncovering the process flows and bottlenecks based on existing IT logs in a bottom-up manner.
Essentially, BI assumes that the underlying processes are known. Process mining takes the stand that even well-defined processes usually don’t go as planned and need to be brought into light objectively.
No. 2: To be able to apply process mining on data-warehoused logs certain requirements need to be fulfilled.
Put simply, to be able to apply process mining techniques, one needs more detailed information than to compute pre-defined KPI dashboards.
Traditionally, data warehouses contained only aggregated data. For example, one would only store one data point for each process instance’s cycle time. In contrast, process mining requires at least one data point for each activity in the process and must keep track of the different process instances.
With an increased focus on continuous monitoring and advancements in data management technology, there now exist data warehouses that hold on to all the detailed, “raw” data points that are a prerequisite for process mining. In this case, process mining can be used as a complementary analysis tool on top of the data warehouse. Otherwise, more direct, native data extraction mechanisms need to be employed.
Makes sense? Please join the discussion and let me know what you think!