Professor Wil van der Aalst is widely considered the founder – or ‘Godfather’ – of process mining and was the first to establish the term across academic and business fields. He started work on process mining in the late 1990s, disappointed with how ineffective workflow management technology was at the time, and with a desire to help organizations both understand and improve their processes using data.
Fast forward 20-something years and Van der Aalst is now Chief Scientist at Celonis, the leading process mining vendor in the market, and he is thinking about how the technology can be pushed even further to provide organizations a more ‘three dimensional’ view of how their process work, where they interact with each other, and where inefficiencies lie.
This latest development is being coined as ‘object centric process mining’ (OCPM), which aims to allow people to map beyond single processes and to not have to go back to the data every time they need another view of that process. The best way to think about it is that whilst process mining is very effective in following a single object through an organization (such as an ‘order), it fails to take into account other objects that impact its progress (such as production, shipments, procurement etc).
Understanding how all of these ‘objects’ interact with each other, in real time, can allow an organization to fine-tune operations in a way that previously would have required manual workarounds and some inefficient guesswork. OCPM – which underpins Celonis’ new Process Sphere technology – maps how all these moving parts intertwine.
If a patient presents to a doctor with a sore right arm, the doctor doesn’t just evaluate the arm, but will look at other areas of the body to understand what is causing the symptoms. A clumsy metaphor, perhaps, but the key thing to recognize is that these complex organisms (organizations, or the body!) don’t operate in silos – everything is connected.
Speaking with diginomica, Van der Aalst says:
If you look at classical process mining, the focus is on the single process. You would like to analyze one process: for example, how is an order being handled? You can think of an order as an object that is flowing through the organization. Several people are working on it, until it is
That view works very well if you focus on a single object. An order being handled, how a patient is handled at the hospital, these types of things. The focus is on the single object and it makes life very easy. You just collect these events related to the one object and then automatically generate the process.
The metaphor that we often use is that you can think of that as a 2D picture, right? You look at a single process, from a single angle, and that’s it. There is nothing wrong with that. Many thousands of organizations are using that successfully every day.
But the challenge that you find, if you think of it as a 2D picture, is the moment that you would like to look at your organization from another angle, you need to get back to the data. So data is always being extracted.
And it’s this last requirement that makes traditional process mining somewhat inefficient. As Van der Aalst notes, it works and it’s effective, but the moment that you want to view the process in another light, or change the angle, you have to go back to the data and do another extraction.
Another limitation is that because you’re following one ‘object’ (again, best to think of this as one process, such as an ‘order’), the assumption is that everything that happens is related to that single object. But, as we know, most activities that take place involve multiple objects.
OCPM is essentially process modeling that allows for multiple objects to be involved. Van der Aalst explains:
That sounds very abstract. But most of the activities that take place involve multiple objects. Think, for example, if you order something from Amazon…you go to the website, you order several things, and then you push the ‘buy’ button, right? I’m going to pay, I’m going to order this. Then that is one event.
In that event there could be three different items I want to buy, for example. So I have this order, I also have a payment, and at least three items. And it could be that of these three items I want to buy, two are in stock.
And the two items that are in stock, they will be delivered in a package. But there’s one item that was not in stock, and it comes only in two weeks time in a separate package. This is the reality.
The reality is there are different objects interacting with each other.
Finding the value
Van der Aalst says that OCPM allows organizations to “view reality as it really is”. In the traditional sense, you would look at it from the angle of an order, from the angle of a payment – but not in terms of, how does all this fit together? He adds:
You get this three dimensional setting where you simply look at what has really happened – events and objects interacting with each other. Then you never need to go back to your data again. Another metaphor that you can use is the comparison between a static X-Ray and an MRI.
What does this mean for organizations that now have this three dimensional approach? It’s not that the 2D analysis was ever wrong, but OPCM is very efficient because it allows organizations to store the events as they take place. Van der Aalst says:
It’s very efficient, both in the extraction and storage. You extract the data only once and then you can pick your view. So it is enabling organizations to use process mining in a much broader way where they do not have to go back to the data. Not that they could not do it before, but before this would take much more effort.
And it’s not just efficiency, the real value comes in understanding how processes work together. Looking at Celonis’ new Process Sphere technology is a good way to visualize how this looks in practice. The technology essentially provides a subway map of a variety of different processes working together:
Explaining the impact for organizations, Van der Aalst says:
This is providing you insight. If you think about my earlier example of placing an order consisting of a number of items, where some of these items are in stock, or some need to be produced…
If you look at one object, you will not see, for example, when you’re handling a sales order, how that is being delayed by things in production. Using object centric process mining, you bring this together in a very flexible way, and you can basically choose your angle that clearly shows you in the sales process how we have this problem – but it’s not caused by the sales department ,it’s actually the production department.
Or if you have an excessive number of shipments, it doesn’t help to just look at the shipments, you want to see: what is the reason that individual items are shipped separately? The only way to understand that is if you can see their relation to other objects.
My take
It’s early days for OCPM, but in theory the technology holds promise and could change the game for organizations that are thinking about process reengineering and process mining technologies. Understanding processes in isolation is one thing, but mapping them across an organization and seeing how they interact with each other is another. Companies know that there are swathes of inefficiencies across their functions, but getting insight into how one minor process impacts another way down the line is hard. What we need now are the use cases to showcase the art of the possible with OCPM – because I would wager a bet that many of the results will be unexpected and surprising.
