Decoding Destination Appeal: Multimodal Social Media Videos in Domestic Tourism Forecasting

Abstract

Social media videos have become an increasingly important influence on tourist behavior in tourism and hospitality research. However, most existing forecasting studies have relied primarily on structured video metadata, while making limited use of multimodal content. Grounded in dual coding and signaling theory, this study systematically extracts and quantifies visual aesthetics, technical quality, and audio loudness from official tourism videos. We use daily visitor data from two tourist destinations, namely, the Forbidden City and Jiuzhaigou, and short videos officially posted on Douyin to evaluate the forecasting performance of video features across multiple forecasting models and scenarios. Results show that video content features, especially visual quality and loudness, outperform user interaction metrics such as likes in terms of accuracy and robustness. These findings highlight the value of multimodal video content in tourism forecasting and suggest that destination marketers should better align digital content strategies in an increasingly video-driven media environment.