Predictive Modeling of Early Dormancy in New Digital Banking Relationships in Morocco: Contribution of Open Banking in Reducing Customer Inactivity[La modélisation prédictive du gel précoce des nouvelles entrées en relation de la banque digitale au Maroc : Apport de l'open banking dans la réduction de l'inactivité client]

This research examines the phenomenon of early freezing of digital bank accounts in Morocco, characterized by significant inactivity in the first three months following account opening. Faced with the challenges of profitability and customer satisfaction that this phenomenon raises, the study proposes a predictive model integrating data from open banking to identify and prevent early inactivity. Through a mixed methodology combining retrospective analysis of transactional data and longitudinal monitoring of a cohort of 150 new customers of a Moroccan digital bank, the research characterizes at-risk profiles and evaluates the marginal contribution of multi-banking data in improving predictive capabilities. Three modeling approaches are compared: a reference model using only the bank's internal data, a model enriched by open banking, and an advanced hybrid model integrating deep learning techniques. The results reveal that the integration of open banking data improves predictive accuracy by 16.4%, offering Moroccan digital financial institutions operational tools to strengthen customer engagement from the early stages of the banking relationship.