Proactive prediction of pre-stay hotel service failures: Evidence from longitudinal field data across major holidays

Much of the existing literature examines hotel service failures only after they have happened. To enable proactive prediction and prevention, this study extends Signal Detection Theory by incorporating advanced machine learning algorithms and proposes an innovative Proactive Service Failure Prediction Model to reduce failures during peak periods. In collaboration with Trip.com, we collected an extensive longitudinal dataset covering an average of 37,883 hotels across major holidays from 2022 to 2024 to rigorously evaluate our proposed model. Results from the field study demonstrate the effectiveness of our approach in identifying hotels at risk of service failure during high-demand times. This research contributes to the literature on service failure prediction while providing a practical solution for hotel service management. By shifting the focus to proactive measures, this work contributes to the hospitality industry better anticipating and mitigating service disruptions.