About Ponies and Unicorns
People often ask me: “isn’t Process Mining just like a specialized sort of BI?”.
Though a common question, to me, it sounds just as weird as: “isn’t a unicorn just a type of pony?”.
Well ponies and unicorns do look alike, just like BI and Process Mining do. But they are all different beasts.
I won’t go over the difference between ponies and unicorns… I will leave that to my young daughter.
So, let’s focus on BI and Process Mining
They certainly share their resemblances:
- Both are analytical tools that try to depict reality from underlying data
- Both usually have capabilities to slice and dice data, to help correlate information and drill-down to specifics
- Both tend to be graphical and flirt with or use machine learning/AI for more advanced analysis
However, two fundamental things tell Process Mining apart from BI:
- A predefined model: in process mining data takes the format of events, which usually present at least 3 pieces of information: a timestamp, a case id and an activity name. The mere existence of these pieces of info allows process mining to perform functionalities such as Play-In (discovery), Play-out (simulation) and Replay. They also enable inferring process models, performing automated conformance analysis and process simulation. All these things are rarely found in BI solutions.
- End-to-End visibility: the very philosophy of entangling events into cases lead to an end-to-end view of processes. This avoids partial, sometimes biased, analysis of data, that don’t tell all the story about the underlying processes nor the context of the business. It is this inherent end-to-end property found in process mining that makes it so appealing to automation and digital transformation projects.
Does that mean process mining can do anything a BI does and more? Not exactly. If the input data does not look like events, it is very likely that a BI tool will outperform Process Mining in its analysis. Time series oriented data, for example, are hardly Process Mining material, unless some pre-processing is performed to turn it into events or to enrich existing events in other datasets.
All that being said, it is fair to say that Process Mining is more than a BI with a magical horn on its front head. They are different animals and do eventually complement each other.
As a rule of thumb, if your dataset is based on events, and you need to analyze/manage a process, or something that looks like a process (customer journey, for example), you are better served with a Process Mining solution such as EverFlow.