Counterfactual Explanability: An Application of Causal Inference in a Financial Sector Delivery Business Process

Service owners, administrators, and business process analysts are constantly confronted with a dynamic of operational changes aimed at aligning business processes with the demands and requirements of the environment. This compels them to take actions that enable the efficient redirection of available efforts and resources. The challenge lies in obtaining a structure of the variables and causal relationships involved in the actual execution of the process, enabling the evaluation and response to potential operational scenarios while minimizing the uncertainty associated with selecting an improvement plan for a specific business process flow. This paper presents a method and its application for evaluating decision-making in the context of business processes, specifically regarding the topological direction and counterfactual explainability of variables that have a causal effect in potential improvement scenarios. This approach is achieved by combining techniques derived from causal discovery and inference, as well as process mining. The technique has been validated through a real-world case in the Chilean financial industry, specifically in a credit card delivery process. During this study, the underlying causal relationships in the operational flow were successfully identified, enabling process managers and analysts to evaluate the causal effect of interventions (counterfactuals) and select the most efficient and goal-aligned improvement actions. A broader application of this approach allows organizations to justify the estimation of the causal effect of an action plan through counterfactual reasoning.