End-user’s corner: Bruna Pereira

end-users corner

An interview with Bruna Pereira

Bruna Pereira is the technical process mining lead at a large Swiss accident insurer. She is responsible for the introduction and firm-wide adoption and provisioning of process mining.

Where and when did you first hear about process mining?

The real first time was during my master's education, but we didn't treat it in depth. About two weeks before I started in my current role, I was looking for a topic for my Master thesis. My boss suggested focusing on Process Mining. Because of this, I started to research the topic and decided to take a closer look at it. For me, this felt like the first time that I heard/read about Process Mining. At last “Process Mining in the Insurance Industry” ended up being the topic of my Master thesis.

How do you use process mining in your organisation?

We use Process Mining in two ways.
The first one is in our internal audit for data analytics and for IT-Audits. Within the IT-audits we use Process Mining in the audit preparation, the execution of audits and in follow-up work of already processed audits.
The second way is to analyze our processes to reveal optimization potentials, contradictions or non-working business rules. Indeed, we tested Process Mining for some of our processes and we are now improving our working ways based on those recommendations. Our aim is to offer a kiosk-like service for our company (our agencies included) that enables employees to send us an order for the analysis of a process. We are convinced that we will achieve a great value-add in this way.

What is the major challenge while using process mining?

One of the challenges is the one mentioned above – natural skepticism. But this is a small challenge and a necessary trait to not overstate the benefits of process mining.
In my opinion, the biggest challenge comes with the data. Specifically, it is to get right and complete data. Enough time should be dedicated to analyze and check if the data is correct, as well as to get the construction of the event log right. Using incorrect data for Process Mining – be it because of incompleteness, wrongly linking cases, or their attributes – can be dangerous, as wrong conclusions can be drawn, discrediting it as a reliable process analytics method. It makes it easier to check the data correctness when the knowledge and understanding of the process are given. Indeed, here mixed teams including business and technical individuals should be preferred over merely technical teams. The biggest effort while using process mining should be used in the analysis of the data, not on their preparation. But we are unfortunately still far from this point.

What was your “ah-ha” moment while using process mining?

Hmm, difficult question. In the beginning, I had lots of “ah-ha” moments. But even if it’s a bit embarrassing, I think that “ah-ha” moment was the first time I uploaded a log file to the tool and saw the result. Concretely, it was the moment I saw with my own eyes how easily the process can be visualized and noticed that there are lots of options to show different values of the process (such as the fastest transaction, the longest transaction, and so on). Seeing is indeed believing.

How do you see process mining in the future?

All the companies are progressing with digitalization. The digital footprints will be more accessible, systems more integrated, process more automated, and therefore the analysis will be – at least in theory – easier. Customers are expecting a fast and easy handling of apps, websites and tools. If you don’t have a platform which is customer-friendly, maybe customers will abort the usage of that platform and opt for a better one, offered by a competitor.

Process Mining helps to improve all the processes running on these platforms, thereby providing a better customer experience. But the optimization cannot be done only with Process Mining. Process Mining will help you a lot to find out where the real weaknesses of a process are, but if not backed with a process improvement methodology, such as six-sigma, and integrated with the improvement lifecycle of a company – best case as part of the management KPIs – its company-wide adoption will not happen.

Also, technically upskilling the backbone of Process Mining by means of new technologies, such as Artificial Intelligence, or traditional data science methods, is a topic of extreme relevance. Research is catching up on this topic, but slower than other communities do. I would very much like to have, e.g., automated predictions and early-warnings on possible deviations and backlogs.