Introduction:
In recent years, the world has seen an exponential growth of data. We often read about how different organizations use our private data, but even those same companies sometimes are at a loss regarding which of their data to harness and where or how to use it. Enter process mining: a set of techniques that, by “digging” the available data, the “outcome” can help to analyze, read, understand, or even optimize the company/team’s workflow.
What is process mining?
Process mining is built upon a mixture of different IT areas, namely machine learning, data analytics and project management. Besides those skills, it also asks for the use of different information systems that collect the data and organize it according to the context in which it is generated. These systems may be, for example, online repositories, workflow management tools, ERP, or CRM systems.
To proceed with a process mining analysis, there is a sequence of steps to follow. Each of them will have a different outcome, and even though some rely on the previous one (as in their input is the output of the previous one), each of them can be used singularly for different purposes.
Process Discovery
Process Discovery is the first and fundamental step of process mining. What it does is exactly what has already been described previously: given a set of data retrieved from information systems used by the company, process discovery uses machine learning, data analytics and process mining algorithms to model a visual representation of the process that, according to the data, is taking place. Given this model, it’s possible to see the different workflows that exist in our process, which paths are the most common, if there is any bottleneck or any problem in the process, or even what opportunities there are to improve and optimize the process even further. In the end, Process Discovery outputs the process modeled as it is being done in real life.
Conformance Checking
Given the model built in the previous step, Conformance Checking is useful to compare the obtained model with the logs used or with a theoretical model that was thought beforehand to identify any relevant deviations. It’s a way to cross information and to make sure that our model is following the paths and the workflows that it is expected to follow. If not, then looking into its visual representation it’s easier to pinpoint what are the most troublesome areas and where changes must be made.
Process Improvement
Process Improvement occurs when, after modeling the process, new data comes in and adjustments must be made to keep it up to date with what is happening in real-time. This step often recurs when, after conformance checking, there’s the case where the real model is more accurate than the theoretical model – this may happen for many reasons, being the most common the fact that in the latter one not all paths nor all the possibilities where taken into account.
By using the logs that were previously employed in the construction of the real model, we can adjust the theoretical one to be more accurate and consider all the workflows that were missing.
Where process mining can be used
Process mining can be used in several different contexts, some more straightforward than others. Following there are some real examples where this kind of analysis is already being applied nowadays.
1. Supply Chain
Companies that have to deal with big supply chains may find some difficulties when managing all the orders and requests as well as the need to track them all until an order is complete. With process mining, by using the logs and the paths that each order takes until it is finished, the model of the supply chain can easily be built to then identify if any bottlenecks are slowing the process or any opportunities to improve the way that these orders are managed.
2. Healthcare
Another similar example is when it comes to applying process mining to healthcare services. Just like the supply chain example, by tracking down the path that a patient followed from the moment it makes a request or is checked in a hospital until the moment it leaves, process mining can be used if there is any moment where the patient was left on hold for too long or if there was any way where the path could’ve been optimized for a better or faster treatment. This chain of thought can also be filtered to see how doctors or nurses are moving between each patient and if there’s any way their work can be facilitated or made easier for them.
3. Banking
Another example regards banking or financial purposes. Process mining can be used to detect fraud in an easier and faster way, as well as ensure regulatory and legal compliances, and helps optimise many of their operations such as loan approval processes. This way bank and financial managers can enhance their risk management and their operational transparency and efficiency in this kind of high-risk systems.
Next Article (part 2) we will analyze the importance of PM in digital transformation and the PM trends for 2024. Stay tuned!
