Some weeks ago I attended the 3rd “Process Mining Camp”, hosted by Fluxicon. As we have been looking at and working with process mining tools for some time now, this event immediately drew our attention.
For years the TU/e of Eindhoven has been looking at how data science evolved, and the “process mining” part has emerged over the last 10 years. With the evolution of a freeware tool to a full commercial product, Disco, the solution developed by Fluxicon, they are ready to take it to the masses.
But let’s face it, Disco is not the only process mining product around, and it was clear throughout the day that process mining as a “science” or way of working is still not seen as an option by the majority of the companies, mainly as it is unknown.
From a distance, it looks perfect: You take all your data, load it into a product and that will show you with a click of a button how your process runs. “MAGIC” one could say.
Frustration point 1 begins already with the “your data” part. Process mining tools need an event log, a step by step list of everything that happened with a “case”. As these wordings are used often in IT support, it felt like every example demonstrated at the event used a helpdesk or call center case study to show how process mining helped them.
But even in the situations demonstrated at the event, users say they use the 80/20 rule, meaning 80 % of the time is spent finding, preparing and loading the data. Only 20 % of the total effort/time is used to actually review the process and look for anomalies.
But not all news is bad. I truly believe that Process mining will be on the list of every auditor/process engineer or data scientist in the next 5 years. Why? Simply because you do not start with a predefined narrow question.
The power of process mining is in the fact you can take 100 % of your data for process A and look at this in a logical way, without the need to ask a simple question other than “how does this process look like in our organization”.
But let’s not be misguided by simplified marketing process illustrations which contain 3-4 simple steps: It was staggering to see how complex even the (in theory) most simple of process becomes once you look at the real process stream. Looking at the business cases of both ING and RABOBANK, it was clear to see that they got bottom line savings just by looking at some processes that are part of their key activity.
Like the migration of data analytics, from using samples to using 100 % of the data I see this happening with these tools. Where in the past one would ask how a process runs and get a personal view of several people that all might have different opinions, process mining allows you to see the truth, as horrible as it can be and without any subjective opinion.
Another fact that surprised me is that every speaker mentioned ERP systems like SAP® are hard to use for process mining, simply because the structure of data is not aimed at looking at the process, but at the individual case. This also explained why none of the cases already used data from the ERP system. However, based on our process mining project experience, this is not a showstopper by default. Data out of SAP® can be prepared easily in a way to be used in process mining solutions.
Prof. van der Aalst (author of the book “Process Mining “), amongst many others, stated that once data availability is solved process mining could take off, as all the data that has been gathered over the years related to the core processes of a company would be available for mining.
So what will the future bring?
It was clear for me that like electric cars there is potential in the Process mining approach. Just like the challenge of batteries for electric cars, once the data (the fuel) is available in a format that can be used the results are for the taking.
The current users, as demonstrated in the cases, are early adaptors. Data extraction and availability of that data is the element that is currently holding them back from doing more. Once this is solved, the investment can be where it should be: not in the 80 % for data preparation and loading, but in the actual reviewing and mining of the process.
So to conclude: For me Process mining is a tool of the future available today. While currently it still needs some tweaking, it might also give you that cutting edge advantage.
I for sure will keep an eye on this technology as it makes data visual and that is clearly the future of data analytics. For comments or questions, feel free to contact me at firstname.lastname@example.org.