Getting a broader view with Object-Centric Process Mining

Back in the days…?

?Whenever I am asked about the most complex process mining project I have ever executed, there is one that immediately comes to my mind. It was for a huge European manufacturer of car spare parts which was just starting with their process mining journey. I was brought there to help finalise the Proof of Value phase. At the time I joined, all the delivery (postponed) deadlines were passed and yet the project was still far from delivering what should have been a straightforward O2C implementation. The client process was not in line with what our wide experience led us to expect.? The gap was the client’s understanding of the Order to Cash process was much wider than that usually applied to the term in process mining. This is typically defined as “follow the sales order until it is closed with invoices”, whilst the client understood “let’s see everything around the sales”.?

When I arrived at the location, I went directly to a Senior IT Project Manager who was responsible for the project client-side. I wanted to check at the source where the problem lay and what her expectations were from us. After first making it clear how important the project was to her personally, as by bringing a revolutionary technology to the company landscape she was trying to build a case for her future promotion, she went on to explain what she expected from mining through the O2C process. By the time she finished her explanation, not only had I already understood where all the troubles were coming from, but also felt like a flip switched in my head! How she drew up her vision of process mining allowed me to look at the topic, especially process discovery, from a new, much wider perspective. As I recently realised, it was the very first time I came across a concept of object-centric process mining.??

The client’s high (and very much justified) expectations of a “revolutionary technology” around what process mining was then offering revealed (also to me), one of the biggest shortcomings of the technology. That shortcoming was only recently addressed by one of the leaders in the industry.?

?

So what is making the difference…?

Traditional process mining is based on the idea of a case (for O2C defined usually at the level of a Sales Order Item) and assumption that each step in the process of handling certain case relates exclusively to that case. Yet obviously very often standard steps (events) in processing of any case, like a sales order item in our example, may relate to something that could also be treated as another type of ’case’, if looked at from a different angle. For the O2C we can imagine ‘delivery’, ‘package’, not to mention ‘invoice’, also being treated as ‘cases’, which could be further investigated by anyone analysing the process to get to the bottom of inefficiencies in the overall process. Such a broad, or rather ‘complete’ look at things is usually very much desired by the business. Looking at our example of O2C, consider a few axes across which businesses can look at O2C and questions they can ask when analysing them:?

• Process of obtaining what is later sold - was it manufactured ad hoc, brought from a different location, or maybe the path that the item goes through is even more complex and that is where delays happen??
• Delivering goods to the client - how many deliveries did it take to close the sales order? By what means was the load transported and were there any unexpected events on the way? Is the way we send things out to our clients optimised from the logistics standpoint???
• Issuing an invoice - were the correct payment terms and tax codes entered by default or was there a need to modify the invoice after it was created? Who is to be accountable for lengthy processing there??
• Cash collection - how costly was it? How many calls needed to be made or mails sent out before money reached our account??
• …. the list goes on…?

?

The need for a new approach?

At the time I was first confronted with such a view on what actual process discovery should look like, there were no tools with right-off-the-bat functionalities to support the idea. Consultants would usually push back on similar requests and propose fragmentation of the wide process into smaller pieces, so it was possible to identify a discrete ‘case’ for each part. In Order To Cash, it would come down to creating separate data models, process maps and analytical dashboards for Orders, AR Invoices, Outbound Deliveries or any other objects included in the process that a client wanted to have a closer look at. Obviously, there are several potential drawbacks connected with such approach. First and foremost was already described above – the inability to grasp a big picture of a process in a single analysis. And what comes with that – the inability to identify inefficiencies living at the intersection points between processes, or even departments. Another issue is the risk of losing the integrity of data used across built solutions, as it actually is best practice to trim data used in each case to decrease the scale of resources needed to process them. (In short – one set of filters would be put over the data pulled from the source system to analyse orders and another one to analyse deliveries and yet another one to look at invoices; same source and process, not 100% overlapping datasets).??

What the client referred to in this article wanted was to map ALL the documents connected with sales of parts to the final customers. For them, O2C processes would usually start at the manufacturing site, go through various storage locations with multiple internal shipments, follow with deliveries to final customers and closed by invoicing and collecting cash. Some of the contracts our client had with their partners would span over years, and a single sales order would include hundreds of separate deliveries. Invoicing was also distributed over time, and due to the complexities in the supply chain, there was a need to analyse documents that are normally part of the P2P process (purchase orders and AP invoices issued for internal sales connected with moving goods from one site to another).?

?

Turning ideas into practice

Such a comprehensive and wide screening across processes and departments is now easily achievable thanks to advances in the science behind process mining. A couple of years prior, Wil van der Aalst, often referred to as the Godfather of Process Mining, started defining the concept of object-centric process mining and, along with his academic peers, worked to build a simple proof of concept (https://www.ocpm.info/ocel.html ).? A few months ago, one of the leading Process Mining vendors revealed a new functionality in their platform which now enables wide industrialisation of the concept. As it is marketed by them, object-centric process mining:?

“… enables the analysis of interrelated objects and events involved in business processes. An object is a representation of something connected to your business, and an event is a representation of something that happened to one or more of those objects. With object-centric process mining, events are related to objects instead of to a single case, making it possible to fully view complex and interacting processes from any angle.”?

Thanks to a shift from case-centric process mining to object-centric process mining (OCPM), business users will now be able to easily switch ‘perspectives’ from which they look at their processes and thus find improvements and value in the areas previously out of sight – such as intersection points of sub-processes and departments. What is more, discrepancies in data used in models will no longer be a problem as implementing OCPM requires not only an integrity within data pulled from a single system but also against other systems that support interconnected processes (e.g. merging ERP and CRM data). Thus, what we receive is a system-agnostic single source of truth for anyone who wants to take a closer look at what is going on in the organisation. To make sure end users are not lost in the multiplicity of objects, events, and relations of both, OCMP comes along with an interactive contingency table that allows for ad-hoc adjustments in scope of information presented on a screen. As a result, every user can easily define their own ‘perspectives’ from which they want to look on the data.??

?

The end that could have been happier

But what about the client at the heart of this story? They eventually got a solution from me somewhat resembling what they initially imagined. Importantly, I was able to fulfil their most crucial requirement of showing all the documents (objects) connected with any Sales Order on a single dashboard screen. When it comes to modelling the process, I went for enriched versions of a few smaller processes and a customisation that made it easier to relate one to the other. I went above and beyond to avoid customer dissatisfaction and finally received sign-off. Would I call it a success though? Certainly not, as it came along with long, late-night hours spent on modelling data, merging tables from two sides of accounting books, building chains of relational tables and multiple other customisations plus the final effect was way below what could be expected for the resources invested. I won’t even bother you with stories about documenting it and handing over to the client’s internal IT, suffice it to say neither side enjoyed the process.?

?

The next episode

Would my client be able to build the solution they had in mind using OCPM on their own? Perhaps not, because it takes even more expertise than traditional process mining to build robust solutions in it. There is still a need to get the right setup, especially on the end of extracting, transforming and integrating data from various systems into a single data model, before any business user can start creating value out of what is uncovered by it. The mere act of looking at the net of objects and events needs to be trained as well. Yet it would still be far easier and cheaper to get this desired overview of end-to-end O2C process using object-centric process mining than it was with traditional process mining. To me, a true process mining enthusiast, the emergence of object-centric process mining opens a new chapter in my professional career and sparks excitement about all the value I can help my clients create with it.?

?

Want to get more insights? Request a live demo with Proservartner.?

?

?

Maciej Olejniczak?

Process Mining Lead?

[email protected] ?