Anticipating Booking Cancellations: Predicting Optimal Interaction Timing to Minimise Loss

Service cancellations pose a significant challenge to industries reliant on reservations, leading to substantial revenue losses when unforeseen. Accurate predictions of these cancellations allow them to prevent cancellations and ultimately enhance revenue management. A novel metric, the Cancellation Risk Index (CRI), was developed to address this issue. The CRI takes daily cancellation predictions for each reservation over its horizon and measures them to determine when a customer is most likely to cancel. This metric recommends a date before businesses should act to prevent a cancellation. A dynamic approach was used in which features may change over time, allowing predictions to be updated daily for the same reservation. Results show that while the CRI can effectively recommend optimal timing for customer interactions, there is still room for improvement in the overall prediction capability for cancellations. Despite this, the metric successfully suggested appropriate dates when cancellations were correctly predicted and should be intervened. This research advances Revenue Management theory by presenting a dynamic predictive model for service cancellations. The CRI offers business practitioners a practical tool to enhance their Revenue Management processes and improve customer retention strategies. By proactively managing potential cancellations, businesses can optimise resource allocation, reduce revenue losses, and improve customer satisfaction.