Answers are grounded in this paper and cite their sources.
Demand and revenue prediction is critical in the hotel industry, where fixed capacity and perishable inventory make profitability highly dependent on the effective alignment of pricing, staffing, and procurement decisions with expected market demand and financial outcomes. Accurate forecasts, therefore, serve as a foundation for service operations, enabling managers to balance cost efficiency with service quality. Online reviews have become a valuable source of information for revenue prediction, as they capture customer perceptions of service quality and influence booking intentions. Beyond numerical ratings, reviews often combine textual narratives and visual content, offering rich multimodal signals about customers’ service experiences. Together, these cues provide more comprehensive insights into consumer evaluations than either modality alone. Yet most existing approaches still rely primarily on numerical indicators or text, overlooking both the visual dimension of reviews and the complementary perspectives available across platforms. This narrow focus reduces predictive accuracy and weakens the operational relevance of current models. To address this gap, this paper uses consumer multimodal reviews as the primary data source and examines hotel revenue prediction through the lens of the consumer purchase decision process. We propose a theoretical framework based on this process to guide the design of our prediction model, thereby enhancing the accuracy of RevPAR predictions. The resulting model, termed the multisource and multimodal interactive neural network (MMINN), leverages cross-platform multimodal reviews to predict hotel RevPAR with greater precision. Our findings show that integrating textual and visual information and capturing their interactions improves prediction accuracy. Incorporating cross-platform reviews further enhances predictive power by using a cross-attention mechanism to fuse multisource features. Moreover, focusing on multigranular customer attention-related factors at both the review and platform levels allows the model to capture differences in the informational contributions of different reviews and platforms, which further strengthens revenue prediction. This research provides actionable insights for hotels by supporting revenue management, staffing, marketing, and investment strategies while enhancing competitiveness and customer satisfaction, thereby generating potential economic value for hotels. It also contributes to the service operations and revenue management literature by demonstrating the value of multisource multimodal data integration for advancing revenue prediction.
No takes yet. Share an insight, caveat, or question.
Mu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: