Process Mining is THE hot technology in our industry these days. But how do you get involved in it, and which skills do you need to bring to the table?

In this Shorts episode we are talking about:

  • What is Process Mining?
  • How is a Process Mining project different from a “regular BPM project”?
  • What skills and knowledge do you need?
  • Where to go next?

Please reach out to us by either sending an email to hello@whatsyourbaseline.com or leaving us a voice message by clicking here.

Additional information

These are recommended books to get started with Process Mining:

  • “Your first steps into the world of Process Mining”, a short eBook by Software AG. Download it here.
  • “Successful Process Improvement: A Practice-Based Method To Embed Process Mining In Enterprises” by Erik-Jan van der Linden. Kindle edition here.
  • “Process Mining: Data Science in Action” by Wil van der Aalst. A bit academic, but a good start to see the logic behind the apps – Kindle edition here.

Transcript

(The transcript is auto-generated and was slightly edited for clarity)

Roland: Hey J-M welcome to a new What’s your Baseline shorts and to be honest, it feels like it was just minutes ago when we recorded the last one isn’t that isn’t it weird.

J-M: Don’t give away our secrets, Roland. We record these all on different days with the same beard cut and shirt (I’m kidding).

Roland: Yeah, different day, same shirt. No no.

J-M: Well, this one’s a new one all about process mining now I know we’ve talked about it on the podcast a little bit but that’s nebulous about the concepts of process mining. This one’s gonna be really easy – just how to get into it. Because many people are trying to take their first steps, and I know you were at a conference recently. You asked who’s doing process mining and very few people actually put their hands up, but it’s everywhere. People are talking about it all over the place. So how do you get started? So first and foremost, well and let’s just define the term let’s take two seconds to just refresh people. What is process mining, what does it look like, and how is it different from traditional BI?

Roland: Process mining is a process improvement methodology that is using data / evidence-based analysis and decision making.

J-M: Oooh, ‘evidence-based decision making’. 

Roland: Long story short, what you do is you take log files from your systems or if you do fancy task mining you record what your users do and then bring it in but based on that data you recreate processes. And all the processed mining softwares have some visualization in a graph you know like a process explorer where then see all the nooks and crannies what your process do and and how people do rework and how long it takes and what the frequency is and how many processes get lost in that big Spaghetti Diagram and and all those wonderful things. So it helps you to analyze the process based on the data, not on what your SMEs that you might have collected in a conference room tell you what the process is. And I think this is what makes it so interesting for all the people because it’s all the new stuff and it’s evidence based.

J-M: Well, it’s one of those things that I think a lot of people struggle with when they think about designing processes. They think of just ‘as-is’ and ‘to-be’. What’s my ‘as-is’, what’s my ‘to-be’ or my what-if states or my transition states. This is one level below the as-is – this is actually the as-executed. You’ve designed a process the way you wanted it to happen, but your as-executed is something like it (probably) but also quite unlike it. And in some ways process mining actually gives you too much information because it shows you all of the different possible paths and all the variations and what you’re trying to see as a forest. That’s why finding a way to granularity slider is a powerful tool.

Roland: I disagree with you. This is the ‘as-is’ and the ‘as-is’ is messy. So your process model is the abstraction. But maybe to take a step back, to come back to your question – what’s the difference to traditional BI? Because you also look at KPIs and all these types of things. But the difference is you put it in the context of a process you’re looking for. How do I get from step A to B to C to D, and I think this is different from what you would do if you would fire up your Tableau or your whatever click or visualization software you use. Even though you can build dashboards and process mining tools as well.

J-M: Yeah. I just want to loop back a little bit on when you said it this is your as-is. It is, kind of. One of the things that we do with process mining that we miss is often the intermediate steps. We miss a lot of the manual tasks, things that you just won’t capture even in task mining. Things that are done away from a computer, things that are done in human interactions, facilitating sessions that are happening for conversations, decision making that goes on that doesn’t happen in systems. It also doesn’t happen on desktops. It happens in the real world and in person back in the days before COVID So there is a level of abstraction that you talked about above as-executed. But you’re right. It is as messy, messy and as ugly as it is. It is your as-is, kind of, and that maybe is uncomfortable to talk about, but it’s a conversation you need to have.

Roland: But that’s the whole purpose of process mining: to see what’s actually going on. But having said that, just to put that out of the conversation. All those other process improvement methodologies that you’ve done before – like having workshops like value stream mapping like whatever lean, they’re all valid and it’s not one size fits all. So pick the right methodology for the problem that you have.

J-M: Yeah, think about it like you’re still process soldiers going to war, but you’re just giving them better weapons. You’re giving them more information and better insight. These are tools to use for professionals to do even more with what they already know how to do. You know what I mean?

Roland: So, J-M, how is then a process mining project different from, say, a regular BPM project where you want to show whatever regulatory compliance or or improve your process?

J-M: That’s a good question. I think that the big difference that I see in a lot of those projects is where you call the source of truth / the canonical source of truth. Oftentimes traditional BPM projects are a lot of interviews. A lot of best practice knowledge comes from SIs and consultancies and things like that. Whereas in process mining projects, ‘Data is King’. It is. Whatever the data says. is the way it was. Now I always have the expression ‘context is its crown’. So once again, you’re using process mining as a professional tool for BPM professionals. So data is the root of the analysis and the root of the conversation. And the other way that process mining projects look a little bit different from BPM projects is that a lot of process mining vendor technology and their methodologies and practices and their approach and thought leadership actually goes into the way the analysis comes out. So a lot of times, there are preconfigured analyses that kind of give things people tend to want upfront. So your BPM project is kind of accelerated because you’re not developing a lot of these analyses. You’re using benchmark style analyses from the vendor itself to draw on, and then you add custom analyses on top of that. Have you seen that sort of thing?

Roland: Oh yeah. That is for example, if you want to do a conformance check. If you want to upload a process model which is the ‘how should people do the process’ and then the process mining tool compares the data set underneath with that process and tells you exactly where you went astray. Like in a procure-to-pay scenario, the process should not start with you receiving an invoice. It should start with a purchase rec or a purchase order, and all these things. 

J-M: So put yourself in my shoes, Roland. Pretend I’ve never done process mining before I’m brand new. Let’s say I’m a new grad or I’m somebody who’s young in a company trying to explore new technologies. What skills do I need to start the conversation around process mining for me, for my colleagues, for my organization, and for my whole company as a larger conversation. What’s the first thing?

Roland: So maybe the very first thing is you should be a nice person because at the end of the day you deal with people. Even if J-M says data is king, at the end of the day, you’re talking with clients, right? So listen to them, and try to understand what they want to accomplish.

J-M: That’s true. You know, that was what James Denning said – you have to be able to look at yourself in the mirror in the morning. Yeah, we want to hire good people. I love that message.

Roland: All those process mining podcasts I’ve listened to, even though they claim to be very technical. They’re just circling around the topic of ‘oh it’s the people’, good or bad. But anyway, that would be the first thing. The second skill that I would highly recommend developing is that you need to understand what data looks like. And you must be comfortable working with data. You know, if somebody throws you some SAP tables over the fence (which they shouldn’t do), but if that happens you should understand what you see and how those tables are related to each other. The third one is that you need to understand the process underneath. So if you think about a P2P process (procure to pay), you must know in advance the purchase rec comes before the purchase order that might come before the invoice that might come before the goods receipt, and so on and so forth, right. You need to understand how that data is retrieved from the systems, in this example, an SAP system or an Oracle. Which tables is the data stored in? And then you need to be able to take those tables and transform them into a format that your process mining tool understands.

J-M: Yeah, you might want to become familiar with a concept like having a data dictionary. You might want to become familiar with creating your own data dictionary and having those facilitated interviews with your data owners to say “what does this field mean”. So becoming familiar with that practice and then developing templates that you can send out to people is important. So that the next time you get data unfortunately thrown over the wall, at least it comes with some context you can actually use to analyze.

Roland: That’s true, and then that’s the first step. That’s all project hygiene, if you will. The more interesting stuff is about how you need to know what to look for. What are the KPIs? Sticking with that example, when you have an order process, what are you looking for? Is it just generic? How long does it take, or how many of those orders do we have? That’s one thing, right? I think you need to drill down a little bit more into industry-specific KPIs. So, for example, OTIF – ‘orders on time and in full’, is one of the supply chain KPIs that the Supply Chain Council (or whatever their current name is) will be happy to tell you that’s something you should look at. So I think you need to do your homework when you’ve identified the process that you’re looking at or which is in scope, you need to figure out what typical KPIs are to look at. And just going in there without having done that homework – you might get a suboptimal result from it.

J-M: I don’t want to stress that there’s so much training and experience necessary on this one, but a fundamental understanding of statistics might be really helpful for you. So I come from an engineering background. I’ve taken a couple of statistics courses. I know enough about them to have some really basic terminology down. But you don’t need to be a deep statistics expert – just know the kinds of statistical analyses that can be done at a very high level. You don’t necessarily need a data science background. In fact, if you’re working with a good vendor, you won’t have to have anyone with that deep of a background to get some good process finding insights out. Just a basic knowledge of statistics and statistical methods.

Roland: I agree. For example, you should know what the difference is between the average and the median and how that relates to if you have the full data set or just a sample of it. Because that might be two different interpretations or four different interpretations that you might get when you apply this.

J-M: And then alongside that is you want to be able to present information. I talk to people all the time about this. You want to be able to tell a story. Essentially you’re writing a narrative. What does our process tell us? What happened when, and where do we emerge out some opportunities for improvement? What kinds of opportunities are there? So know how to tell that story.

Roland: Yes, but before you tell the story, you need to know what you want to know. What’s my hypothesis before I go in there? Is it: we think this process stinks because Vendor X/Y/Z doesn’t do its job. Or when you think about IT Service management, you know, is it always stuck in that outsourced incident group that’s working on it? And how do they work? So you should have a gut feeling. And then you look into your data and you create those visualizations, and maybe you see a story evolving in your data. And I think that’s a completely different skill set in itself that I know that we’re gonna talk about when we talk about building custom dashboards and all that stuff. 

J-M: Yeah interface design, user experience for design for visualizations. Yeah, we’ll go into that later. But, Roland, I want to summarize this all into something. Really, what we’re telling you is that you can do this! The barrier to entry is incredibly low and there are ways that you can work with vendors and consultancies that can help you get started with almost no information. It only needs a good canonical data set you understand, good hypotheses you want to interrogate, and a good way of telling and selling that story to a larger community. But Roland, where do I go next? What’s the information that I can take away from this podcast, and that’s generally available and oftentimes for free that we could consume to really dig into this?

Roland: So if you’re a complete newbie and you haven’t heard about process mining, there are really good books that are good introductions. And I’m not talking about the academic literature that explains how algorithms work and all that stuff. But it’s written for the regular business person and I will put some links into the show notes. Once you’ve done that and you’ve got a good understanding of what process mining is and how it works, and maybe you’ve watched a couple of videos on YouTube about this, and seen what the different vendors have to offer. Well, then you might want to take a step back and have a look at your organization. Find a problem that they have. And when you do this keep in mind process mining is not for any problem. It’s for processes. It will not help you if you have a branding problem or a market perception problem or a production problem because your machines always die.

J-M: There are other other fixes for those.

Roland: It’s about how you orchestrate processes, and how you can get better. That’s what we want. So once you have that, J-M, what would the next step be?

J-M: First and foremost, you can start looking at actual technology. You can actually go and pilot out the tools. We had a Short about POCs and POVs – just start using the tools. There’s a lot of free downloads / free versions of process mining software. You can download it, start bringing data in (even just dummy data – a lot of those free versions actually have dummy data to download), and get free training on them. So go educate yourself on the practice of using the platform to do things, and start communicating that out with your peers and with your management. Because if you can start to show pretty substantial gains in a very small amount of effort, you have a really good case to take this to the next level.

Roland: I would also highly recommend you take a look at online communities. For example, that little community around What’s Your Baseline on LinkedIn.

J-M: [feigns surprise] Oh my goodness, tell us about that, Roland!

Roland: Well, I don’t want to do the pitch here. But do look around. There are people out there who have the same interest as you – start the conversation with them. Go to virtual user groups, go to dedicated process mining events (once the whole COVID situation improves). So do this. Mingle and become part of the community and you will see vendors also supporting this. It’s not that they just go and say “hey, buy our software and then bye bye”. They definitely have an interest in having a conversation with you. Well, anything else, J-M?

J-M: Not really, Roland. I think this was an excellent place to get started. Thank you so much. You did mention the ‘show notes’ on https://www.whatsyourbaseline.com. We’ll have all the show notes with some examples of things you should read. Roland always has lots of great advice! But this has been a fantastic discussion, my friend. Thank you so much for helping folks dip a toe into process mining.

Roland: Well, then, as always, J-M, glad talking to you. My name is Roland Woldt.

J-M: Glad talking to you – my name is J-M Erlendson

Both, in somewhat unison: And we’ll see you in the next one.