#78 Identifying Process Bottlenecks, Process Mining and more

Hello and welcome back to another episode of the Process and Automation Podcast with the Automation Guys. When you potentially have a problem, or when you identify, or you feel like there's a problem with your— with speed in your business and in business processes, you very often don't know exactly where that specific productivity slowdown might happen. And yeah, if you don't know that specifically, yeah, frustration builds up for within the teams and also within sort of the management teams. And yeah, and it's not just a very, very individual issue with your company. It's generally widespread issue for enterprises, especially enterprises trying to accelerate their pace and increase capacity for work in the, yeah, pursuit to become more digital, to transform the business. And yeah, I want to generally improve how everything runs, customer onboarding for example, or to be faster fasten the market with some, some changes in innovation. So, Arno, I think our listeners might want to know how to identify these bottlenecks.

Yeah, I guess, Sascha, this is the big question, you know, how do you effectively go out and identify bottlenecks in a process or in your processes? And again, you know, if you look at organizations today, they have a very large portfolio of simple and complicated processes. And, you know, these comprise of effectively multiple workflow stages and activities where workflow from one person to another person. You know, just as an example, if you look at managing a multi-tier supply chain to support manufacturing operations, you know, that, that is a very good example where, you know, you can have issues like data silos appearing, and you have to deal with legacy systems. You know, part of those processes within these legacy systems, for example, or within the data silos, you know, can be very hard to uncover, and, you know, it's very hard to reveal them. So to fix sort of different types of bottlenecks that might exist in those processes, you must have a, I I guess, a very clear and comprehensive view of all of the activities that takes place with inside those processes. You know, manufacturing is just one example, but, you know, we covered a couple of examples, lots of use cases last in our last episode when we look at the, you know, process automation within in the insurance industry, for example. So I guess, you know, the point here is You know, you have to sort of judge where changes and optimizations in those processes will have the most impact because effectively that's the area that you want to focus on. And, you know, modern tools like process mining definitely helps you uncover those very, very quickly. So it kind of effectively accelerates that process, and it helps leaders to identify process bottlenecks and gain better insights into how the processes are performing, and that will help you to take corrective actions. So process mining is a very, very important part, in my view, of, you know, the sort of broader process automation and if you look at hyper automation from a strategic level, process mining is a really important part of that. And improving these processes will obviously make them more resilient and compliant and reliable. So I think the point here is that effectively the sort of the go-to, thing to do is to look at process mining because that's going to reveal a lot about these processes and ultimately, you know, the bottlenecks that exist inside them.

So let's dive a little bit deeper into that identifying of bottlenecks. So when we look at the processes itself, so organizations, departments, So they struggle with both short-term bottlenecks that cause temporary setbacks, and there are the long-term bottlenecks that again and again slow down the work within those departments or processes. So let's focus to start with on the long-term bottlenecks, because they are, as we mentioned, they happen again and it produces a constant setback here. So As Arno, you mentioned earlier, so the analysis will help to find the sort of the root cause of the process bottlenecks and the process inefficiencies. So process mining, you mentioned earlier, so it's available for everyone. So if you want to listen back on some of our process mining episodes, yeah, I invite you to go back into the list. We will link as well in this episode here, a few process mining episodes. But so yeah, the process mining tools you have out there will bring that transparency. And then very often, or most cases these days actually, delivering that visual representation of the entire process and different variations of that process and really highlight the areas that need optimization, providing this sort of really the insights. It tells you exactly at what step in the process things are going wrong, or again and again in this department are going wrong. Or take very long. So with those insights, yeah, you will find out what the path for that corrective action might be, and that will then— you will do that again and again, lead sort of that continuous improvement we all want to have in our organization. So, and but how does it work maybe in in detail. So the process mining tools very often look at these event logs. Arno mentioned there are lots of different systems out there. And so yeah, these event logs in these systems tell us pretty much what happened when in these systems. And yeah, by analyzing these event logs across the whole organization, across all the systems, we will have really, really good view of how the process looks like. So we have these timestamps, we know exactly what order number was involved, for example, and that will paint us this picture, the visual representation, very nicely. And of course, yeah, we wouldn't be able to do that manually, that would be crazy. So yeah, this is why we're using the process mining tools. And yeah, Anna, do you want to continue a little bit on, um, yes, what the exact steps might be?

Yeah, so effectively, let's examine the underlying steps to take effectively when you using process mining and the techniques to spot bottlenecks. So the first thing you need to do is really to define what to measure and why you want to measure it. So firstly, you want to make sure you have process data available and These include those timestamps, event log, that event log details that you spoke about. You need obviously to have support from management and sufficient resources in order for a successful execution of a project like this. You need to obviously connect with the relevant stakeholders like the process and data owners, underlying sort of system experts, and also the decision makers. So, you know, now that you can kind of specify the goals of your analysis, you know, at the next point, you need to consider creating a reference or a target process model, which will then be optimised or represent an optimised version of the process that represents your ideal state. And then, you know, you'll compare this with the actual process once it's mapped. And then the next step is to dig into the data. So the process mining tool obviously extracts process data from event logs, and it will categorize the data according to unique identifiers, for example, an order number or a customer number or incident number. And each activity taking place, you know, within a specific process will have the same identifier. So, so it's a means to correlate things. Activities will also have things like descriptions and, and, and other timestamps. So again, once the data has been categorised and properly formatted, the process mining tool can then draw a visualisation that maps the actual process from start to finish. So effectively, what you're doing is you're pointing it at your kind of large datasets. Process mining tool will evaluate that, and it will then extract the underlying model, you know, that the process that will represent the process. And then effectively, the third thing you want to do is analyze and evaluate the results. And this is sort of the time that you use, you know, kind of visualization to inform you, you know, what the next best actions are. So, I don't know, Sascha, if you want to sort of just comment on those ones.

So, yeah, once we have all that data prepared, analyzing it is now quite straightforward. And then the first things we might look at is the conformance. Performance checking. And yeah, we have that reference model Arno mentioned earlier. If we have that in place, yeah, you can now compare it to the process, how the process is really running, and that immediately will identify the areas which need that optimization. So for example, you may find that people are skipping or duplicating some activities or executing unplanned activities, or Yeah, that's something what we don't want, are doing a lot of rework due to human error, or that PO number is missing again and again and again. So we have to go back in and request that. So sometimes, yeah, that can add days to the overall, to the running process overall. So the other thing we can now do when we evaluating the results is the model enhancement. So We can use that information we have gathered about the, the actual process. Now we can think about extending or enhancing the process to achieve a better target process. So yeah, this may uncover further bottlenecks and unplanned process sequences that we can get rid of from the model to make it make it a better representation of the perfect ideal target process. So the next one is the root cause analysis. So looking through the process variations we have, so we're thinking about the large organizations, they have 10,000s of, let's say, purchase orders and instances of that purchase order process. So now you can look at these variations in detail and you can help identify where actually the problems always occurs. Maybe this was a specific country, Maybe it was a specific department, maybe it was a specific supplier or supplier group. So if you can figure that out, you immediately know where the problem is. Maybe it is a problem you need to get into with further process automation, with other workflows, workflow tools. But yeah, so with that bottlenecks identified, you can really work on that targeted plan for the optimization. Yeah, and as I mentioned, in some cases you don't even need technology for that. So if it's a supplier, you maybe need to speak to the supplier instead of just building lots of logic and algorithms around it. But yeah, it definitely will highlight everything what needs to be done, either technical or non-technical.

Yeah, and I guess with a final point here is that, you know, once you have identified the bottlenecks, you know, you can actually create a plan for optimization and, you know, you can create dashboards that the process mining tool provides you and all of its reporting features to monitor the ongoing performance against some KPIs. And of course, KPIs or key performance indicators are, you know, quantifiable metrics that help you to measure the you know, performance of a process during various stages. And, you know, the monitoring of these will really allow you to identify areas that consistently fall below the desired performance levels. And, you know, and that will give you a very good clue where your potential bottlenecks are. So, you know, you want to look at stuff like cycle times and throughput time, how resources are utilized. I think that's quite a big one because, you know, it will assess the extent on which, I don't know, your equipment and your personnel are being used to, you know, within specific stages in your process. You know, higher resource utilization indicates a potential bottleneck due to the overburdened resources, while low utilization will tell you that you've got underutilized resources. So I guess it's, you know, it's important to set these targets or benchmarks for each KPI, you know, based on the sort of your desired performance. And really the key is to constantly monitor them and, you know, and to do regular review on these numbers and these figures. And effectively, you know, that's when you sort of can make data-driven decisions to improve your your process efficiency.

Yeah, also the nice thing here was having such a tool or platform in place, a process mining platform, is so when these different variations occur, you can either long-term, you fix it with maybe some of the other processes, but generally you can combine it with, let's say, RPA to do a quick fix in an SAP system or so. Yeah, so this is how you can basically get immediately actionable recommendations nearly, and you can either do them manually or you can further automate them. But that's really, really powerful. And yeah, with that I also mean that it might require just combination of tools, not just maybe RPA. So you might look into sort of workflow orchestration and other tools. Maybe you want to put a little bit of AI in, in, in terms of that resolution. So, so it's very important to look for maybe a platform which can do all of it. So, but yeah, there are so many different technologies out there and platforms out there, but it's not just usually just one thing you look at. So they're probably a combination of things you, you want to pick from to get the perfect outcome. After you gained all your insights and the reasons for the bottleneck. And yeah, for everyone who listens to this episode and is keen to learn more on how to successfully implement a process automation strategy and process mining tools and RPA tools, all these things we cover here on the podcast. So yeah, For everyone who is interested to learn more, yeah, join our theautomationguys.net community. And yeah, we look forward to seeing you there. And yeah, of course, we will be back with another episode very soon. And until then, let's automate it. Unfortunately, that's it again with this episode of the Process and Automation Podcast. Podcast. If you like this episode, please give us a 5-star rating and don't forget to subscribe to this podcast so you don't miss any upcoming episode. We hope you will tune in next time, and until then, let's automate it!