The story of the Godfather of Process Mining w/Prof. Wil van der Aalst
Today’s guest — Prof. Wil Van der Aalst
Process Mining & Process Intelligence
Chief Scientist of Celonis
Professor of RWTH Aachen University
Former Assistant Professor at Eindhoven University of Technology (TU/e)
The author and teacher of the Process Mining: Data Science in Action
& many more citations, papers, advisory roles, teaching classes, and books. No need to say, he is also known as The Godfather of Process Mining
About the Episode
In today's episode, it was my absolute pleasure and honor to host — the one and only — the Godfather of Process Mining. Prof.dr.ir. Wil van der Aalst is the Chief Scientist of Celonis and a full professor at RWTH Aachen University, leading the Process and Data Science (PADS) group. He coined the field of Process Mining, and his research interests include Process Mining, business process and workflow management, process modeling, and analysis. I, myself, took his course Process Mining: Data Science in Action in Coursera from the Eindhoven University of Technology a few years ago, which ignited my passion in the field.
As the professor noted in this episode, not many people and companies, including SAP and IBM, believed at the time what this technology could accomplish. They were surprised but did not implement it. There were skeptics and resistance to change. Many enterprises go back to old ways of doing things because it is easier. They denied the transformation that Process Mining could bring to them.
His students in his Process Mining courses branched out and built the Process Mining companies you see around the world, including; ProcessGold, Celonis, and PAFNow. As the academic field of study transitioned into product companies that aim to sell and bring this unique technology to enterprises around the world, Process Mining became a vital ingredient of the digital transformation initiatives.
Now, set your clocks to 2022 and watch the news — SAP acquires Signavio, Microsoft acquires Minit, UiPath acquires ProcessGold, and Appian acquired LanaLabs.
The acquisitions are a great testament to the customer’s ask to not focus only on the diagnostics but target execution and actions. By combining the rest of the platforms’ automation, execution, and process management capabilities, the enterprises aim to improve their business performance.
Our guest does not need an introduction or marking — The professor is the undisputed Godfather of Process Mining.
Some questions we discussed:
How did you get into Process mining, and what led to people calling you the Godfather of Process Mining?
What was your first inspiration, driver, and reason to purpose Process Mining? What drove you forward?
Live transactional processing and execution management is core to achieving digital transformation. How do you see the process mining evolving over the years not to be only a tool but a product that touches each stage of the process transformation journey?
How do you see Process Mining's intersection with transactional processing within the blockchain? There is a great use case to track, monitor, and report the activities happening in the chain with PM and perhaps run some forecasts to identify fraudulent activities?
The announcement of the Innovation Lab in the Aachen Campus is an excellent example of you pioneering the open-source and accessibility culture for Process intelligence. What are you most excited about this?
Why should you care?
Hear Prof. Wil van der Aalst’s personal story of getting into the field with Petri nets and kick-starting the Process Mining Journey
Better understand the mistakes and lessons learned from the early days, applied to the present and the future of Process Mining technology
Explore new ideations of Process Mining technology’s capabilities like Process Mining for Blockchain
Widen your perspective by listening to the discussion on the new methodologies to build a Process Mining alliance and consensus across the countries and departments in organizations to improve widespread Supply Chain problems globally triggered by COVID-19, Ukraine War, and more.
“Companies are not interested in diagnostics, they are interested in action and the change the technology can bring.”
“20 years ago, the companies were surprised by the technology but did not implement them. Now, they spend billions of dollars to acquire the Process Mining firms.”
As the companies were fascinated by the technology, they were not quickly adopting it. The reason is that software companies typically do what their customers ask them to do, and advisory firms echo the value one customer has for another. If the customers do not ask for a change and don’t know their pain problem — it reaches a stalemate. As disruptive technologies like Process Mining get introduced in the market, adoption of the early users and building the product-market fit is critical.
Notably, the same echo-ing effect as we call it, the network effects, during the early days’ transitions into “If company X is doing it, we should do it too” to stay competitive, innovative, and on top of the curve.
Now, everyone, including the disbelievers, wants to try and adopt Process Mining.
Transcript
Alp Uguray: Welcome to the masters of automation podcast, in today’s episode, we have Prof. Wil van der Aalstessor Aalst, also known as the godfather of process mining. Welcome, Prof. Wil van der Aalst. It’s a pleasure and my honor to speak to you in this episode.
Prof. Wil van der Aalst: My joy to be here!
Alp Uguray: Prof.Wil Van der Aalst is a chief scientist at Celonis and a full professor at RWTH Aachen University, leading the Process and Data Science group. He coined the field of Process Mining, and his research interests include Process Mining, business process and workflow management, process modeling, and analysis. I, myself, took your course Process Mining: Data Science in Action in Coursera from the Eindhoven University of Technology a few years ago, and that kicked off my interest in the field, and my journey with process Mining since then you've been an inspiring figure in process mining is the key ingredient component of the future of world technology stack at a high level, as we imagine and standardize efficient and optimized process in our daily jobs. Process mining takes an essential role in revealing problems and finding a way to pave the problems with data back metrics. Personally, it was a valuable moment for me traditional I want to get started How did you get into Process mining, and what led to people calling you the Godfather of Process Mining?
Well, perhaps to start with the last question, I do not know, right! People have been saying for at least ten years or longer. How does the term get into existence I do not know, of course if you look at the key people in the process mining fields, then, you can see that these are many people that I worked with them in the past. Then to return to the fist part of your question, if you look at my history I like my original research was very much in the field of Petri Nets and building simulation models right! My first applications but mostly on logistics and it was mostly building simulation models using Petri Nets of these processes and then in the early 90s, there was certainly like an incredible attention for work flow management and certainly everybody started to talk about it and the ideas was new. I even told you that I'd be people working on this in the 70s party in the 90s it seemed that it would be a reality right! it would be possible to do that so there was an explosion of vendors that started to sell work flow management products at the basic promise that everybody including myself okay let's just make a model of your business processes and then after you modeled it, you push a button and you generate an information system that is supported in that process. And myself and also many other people believed that work flow management would be a key technology used in any organization. That was the early 90s and mid 90s and then I was involved in many selection processes of work flow management technology and one of the things that I experienced is that many organizations would buy work flow management software, then would try to put into production and often, it was never put really in production and I think one of the primary reasons was that many people discovered that there was a huge gap between modeling a process, like a Power point picture and the actual process with all of its exceptions and strange convocations that they had to handle. When I saw that so many people got disappointed with this I started to think okay let's not start by modeling, lets start by looking at what people are actually doing based on actual facts and that was, let’s say late nineties that I started to do first research projects that basically look at the problem and how we can automatically learn the model of what people really do and in the beginning it was just a scientific problem but for me it was immediately clear that this approach in the end would be much more important than just modeling processes. Because I see the failure of modeling that what people model and what's happening in reality that has not so much do with each other. Everybody in the process mining fields knows the term Spaghetti model. I think I introduced over twenty years ago and that is the process model that you get if you apply process mining which is a real process which is very different from that. So, it was in the late nineties I was the only person working on this. Then, I build up a group like the largest group in the world specializing in doing research in this field and many of my students also started companies. Because, it was immediately clear that I want these ideas to be used and for that they need to be companies that sell this technology. But, it what was a bit anecdotal is probably like 20 years ago I was trying to sell these ideas to SAP, and to IBM and all the tech companies I was trying to add work flow management and I was trying to convince them all and everybody was surprised by what I could do but they didn't implement it and now they are paying millions and billions of euros to small companies that basically do what I was talking about twenty years ago.
Alp Uguray: and they are paying the price for the chances that they did not take at the time definitely. And that is a great because it started a great field, that it started great companies like (…. Process Gold ) and other ones and create an eco-system and then the academic theory behind it applied to the business life and I think that's helped to reiterate the theory over time and improve them and then see how we can resolve the spaghetti processes that they have and then that's very interesting, the meet up point of business and academy and for yourself what was your Inspirations at that time you mentioned that you mentioned people were a bit resistant to adopt it quickly so what made you inspired and drive forward?
Prof. Wil van der Aalst: Yes, it was like a super interesting scientific problem and I still have to fight against ideas. So, many people that hear about process mining they think that this is like data mining or this is like a specific type of machine learning and … etc but, it is not. It is very different. If you look at algorithms that are behind it, they are very different from algorithms that people knew before. So, for example if you look at liquid machine learning which is of course super important at this point and time. Technically, you could say process mining is a machine learning because you learn a model using algorithms so you could define it as a machine-learning problem or as a Nİ problem but in reality the algorithms are very different and one of the things that always fascinated me right from the beginning is that if you have processes that have half sequences, that have choices, that have loops and above all have concurrency and our world is super concurrent and if you have all these phenomena happening all together and the only thing that you see is example traces, example behaviors, and you would like to learn this structure, that is from a scientific point of view is a very interesting problem for over the last twenty five years we booked interesting progress but still have not solved the problem. As so, part of my answer would be it is a fascinating field with fascinating problems. Then, if you look at the adoption in industry, I've always been surprised. So for example, I remember in 2005, that we had we had the fuzzy minor with token-based animation and what we would do is that we would automatically discover the process model and then we would animate the tokens which you cannot find on any system that you would animate it opens and the fuzzy minor that had this very nice, let’s say a thing that it's like the parts of the process that would be very overloaded they would become red blowing , like they were on fire. I would be sitting in rooms, and would take data from organizations and we would show this and people would be astonished right! They would be astonished that it was possible, it looked exciting. Everybody was excited in the room and then I was always thinking ok now they are going to widely adopt this but then I think people often slip back to the normal way of doing things and I think the reason behind it I think the software industry specifically the larger companies they typically do what their customers ask for. And also, many of consultancy firms, they just echo what their customer say. So, if you think about taking advice from consulting firms, they talk to one customer and they echo what they hear from one customer to the next customer. So as long as customers are not asking for it and you don’t really apply it. You don’t have the reinforcing fact. This is dramatically has changed in the couple of years because now half of the 500 companies are using process mining. Here in Germany, probably all the large organizations are using process mining. Now we see that it has this echoing mechanism that was not there in the past. And now that’s why I was surprised why people did not use it and now certainly many people use it and it is still exactly the same technology.
Alp Uguray: And it took some time for the organizations to go back and digest information probably and start to sell it and implementing to the companies. That is very true! In the past where it was and where it is going and now it is very different. The time I was interacting with it, I was installing it on the process and trying to figure out how it works and to conform its checking and at the time I was more playing with it like a data extraction so it was not a live transaction process thing that fits through the API connections and that is one area that I wanted to ask you because a process mining as a tool is actually very powerful because once you connect to the database and have the process graph and data (QQIS) can do call to actions for the bottom next, depending on that transaction and that tools that execution management approach that each transaction can be triggered and something can happen afterwards to result that exception. So, how do you see process mining evolving over the years in and then seeing where it started and where it is that now, where it is growing to?
Prof. Wil van der Aalst: If you see where it started, I think you're completely right! is that it started a bit as a tool for the data scientist right! So, data scientist would extract event data and then it would do the analysis and the results would be a report and the report would be given to management and the management would say based on this and these we would like to change that. So, in the past process mining was seen as a project, just like a process where you would model processes and see where you would improve it, it was like an improvement project. I think what has changed is the following and it also partly explains the success of Celonis company. Celonis did two things different than other companies around that time. One difference is that right from the beginning, they position process mining not as a tool for a data scientist but as tool for people who actually working in the process so that is relatively simple, you need to create dashboards with labels and explanations, pie charts, pie graphs … etc that talk in the language of people involved in that process. So, from a technical point of view, it is not a big step, but you need to make sure that the process mining’s results are seen by many people continuously, right! if you want to approve a process is not like you generate a report and then everything is okay it is a continuous effort to do that. I think that was one thing that Celonis did different than many organizations. What can you see in the last couple of years and you can see that the whole market is going to that direction is that Celonis bought integromat as a means to connect process mining results to triggering automatic work flows to improve the processes. Like, you automatically generate actions to not just generate diagnosis but to interfere in the process. If you now look at the market, for example if you look at IBM bought myinvenio.., ….. ,the process growth was acquired by …Microsoft bought minutes .. and there are lots of mergers in the market, if you analyze these mergers from a distance, what you typically see is the combination automation and intelligence in process mining. I think this is the key idea that organizations do not pay for diagnostics they pay for improvements. It requires that connection and it is clear that the field is adding in that way. Another thing that is important and we haven’t discussed yet is the fact that the extraction of the event data is still one of the biggest topics. To go from hundreds of SAP tables to one dashboard that is showing where problems are is something that still takes a lot of work. This is not computer science problem, from technical point of view it can be done. It is more like how to make process mining somehow domain specific. For certain processes we know what kind of event are being stored, we know typical problems and that explains why many organizations start with purchase to pay and order to cash as the standard processes. It is not because, they are the biggest giant, to be a chief there, right! Because, it is for many organizations the secondary process, it is not the prime purpose of the company. But it is frequently done, because we know how to extract that data, we know what are the reoccurring problems and we can address this. That is the entry level process mining and my hope is that now, if companies have successful experiences with that they will more apply in areas like production, logistics, etc which are the real key processes of many organizations.
Alp Uguray: And that’s a very interesting point because the digital footprint, the event logs, if it is not there, it means it needs to be transformed and it needs to be shaped so the process mining can recognize it. And if the data is not there at all then, you're unable to use software to drive inside. To tying that to another emerging field, blockchain. What are your thoughts about blockchain’s interaction with process mining? Because, it does produce event logs and I know some of the chains do not give the required granularity of information and some do. However, with that live transaction processing and the massive amount of transaction there as well. It can be very interesting to see what insights are happening there? What are your thoughts?
Prof. Wil van der Aalst: So, I'm a bit skeptical to be quite almost right. On the one hand, we also wrote a paper on process mining and blockchain data, but I see the blockchain is a bit similar to something like the cloud, right! So, it is a type of technology that is important and before, process mining mostly are done on the laptops and now, you can see more and more organization they don’t run it in the company anymore they run in the cloud. Because you don’t need to worry about scalability and all of these things to take care of. The same is that I think for certain applications blockchain will become a standard technology. I think it is less relevant than something like moving to the cloud right! I think it is more like a niche type of area . Once you have blockchain data that allow you to extract event data that you may not have in a regular or normal setting. However, just like with clouds, doing process mining in the clouds, it does not fundamentally change the problems right! The fact that, whether it’s a blockchain or a we file or a sequel (SQL) database, once you have done the extraction, it is the business as usual. So, I think from the view point of changing the process mining algorithms or something like that, I think, there is nothing new. At the same time, it would be very interesting to think of applications of process mining and settings that you normally cannot address. And one of the untapped potentials of process mining is what we call let’s say federated process mining and that is process mining is typically is done inside one organization, right! because that organization owns the data. You can do the analysis you can do the improvements, but there are also, look at the supply chain that broke down at times of Covid and Ukraine war, etc. Supplying chains are not one company thing, many companies involved and perhaps blockchain could help to lift process mining to the interorganizational level. That you agree on the standard medium. With the goal that you fix what the interactions are between different companies and that provides a single source of the proof shared between different organizations. So, I think there is an application but I would say it’s more like a niche. I know there many people who are super enthusiastic about blockchain, but you have to wonder what are the applications, why you really need something like that, that is not a trusted partner that just manages all of that right!
Alp Uguray: Yes, I agree with you. An echo that it can be a great opportunity to come as an inter-organizational opportunity so that the event logs can be accessed through one platform and everyone can be tracefully benefit everyone. It's not stocked in someone’s dashboard. So, it’s almost that time. I have more questions to ask, but I know you are very busy, and I really appreciate you taking the time and joining the call and having the podcast with me. It was a great pleasure to learn from your insights. Thank you very much for joining the session
Prof. Wil van der Aalst: It was a pleasure to talk to you. It is always nice to talk to a former student of my course. I think in total now it has been taking over 150000 people that shows the growing interest in this topic. Thank you very much for your questions.
