Imagine for a moment a CFO monitoring different aspects of the business with a scorecard. It would show different indicators such as working capital, productivity, risk, and so on. However, something catches his eye: the working capital metric is in the red because the payment period is much shorter than expected, which means that his company is advancing money to its suppliers beyond what the purchasing process can tolerate. With process mining, you can discover the cause of this deviation, the economic impact to your business and even automate the modification of payment terms and generate informative alerts to correct the deviation.
When trying to find out why this is happening, traditional re-engineering has certain limitations when trying to evaluate internal business processes with the aim of making them more efficient. Detecting and solving pain points, regardless of the methodology used, always starts with a study of the current situation and the determination of the ideal state, the analysis of which sometimes presents a high degree of subjectivity.
However, the process mining methodology, with a strong technological basis, allows to observe and analyze the reality of the different instances and occurrences of the processes in an objective manner, based on their digital footprint. Thus, instead of imagining the reasons, we can observe them thank to process mining.
Data in the banking sector
One of the sectors that can benefit the most from this technology is the financial sector, as it has got a huge amount of customer data. According to Sopra Steria's 'Digital Maturity of Spanish Banking 2023' report, there is increasing interest in digital financial solutions, which represents an almost inexhaustible source of data for institutions.
In this scenario of exponential data growth, the process mining technology has the necessary foundations for its application. It is a discipline that allows us to know the reality in the occurrence of processes based on the automated recording of events and timestamps. The process and all its instances are discovered and traced in their real form, with its iterations, deviations and time recording, which allows us to discover how they work, monitor them and, subsequently, improve them, even automatically.
There are various process mining tools, but in essence, they all use data and AI technologies to recognize and interpret process nodes and reveal unexpected deviations and loops within each stage, how often they occur, actual execution times, rework cycles, automation rates and bottlenecks. The idea behind this is simple: if we are able to see the entire value chain, as well as its variants and the individual steps involved, we can optimize the results.
The real information about the process and its instances is captured in a tool that allows us to mine its operation through a graphical interface, in order to find the occurrences that move it away from the ideal state and the reasons why this happens. In the case at hand, this would involve analyzing when the payment order starts and ends. The information is presented in a dashboard with indicators that could show, for example, that the company pays even before receiving the goods.
Another very interesting application is fraud detection. This technique can find variations in behavior and detect irregularities when a pattern identified as safe shows any unusual change. An intrinsic modification of attributes and parameters in a sequenced chain of events or behavioral changes in a user's login attempts or card payments can signal that malicious login or purchase events are taking place. This technology can also be applied to reducing credit approval times, thereby improving customer perception and avoiding loss of business generated by approval delays.
Contrary to what it might seem, this type of project does not involve a migration of the core business (which can continue to rely on the companies' systems and repositories), but rather the implementation of a new process-oriented data model. The timeframe for a typical project would be around two to three months and would include the following phases:
- Pre-analysis: during the first month, entity/relationship data sources would be identified.
- Analysis and implementation: in the second month all the dashboards would be created and the KPIs to be monitored would be introduced in them.
- Mining and identification of inefficiencies: finally, improvement alternatives would be determined and an action plan would be drawn up.
In summary, process mining is a technology that can help to convert millions of data generated by the processes, with a digital footprint, into actionable information. Companies leading the digital transformation, particularly those in the financial sector, are already making better decisions, reducing costs and improving the experience of their customers and suppliers, because they understand their business behavior in detail.