Unraveling the Complex Effects of Content Congruence on Engagement in Sponsored Tourism Videos: An Interpretable Machine Learning Approach

Sponsored videos have become central to travel influencer marketing, whereas the role of content congruence remains underexplored. Based on habituation-dishabituation theory, this study develops a framework to measure content congruence in sponsored tourism videos and examine its association with audience engagement. Specifically, deep learning techniques were used to extract and evaluate the content congruence of various modality features, and interpretable machine learning methods were employed to reveal their association with audience engagement. The results show that different multimodal features and degrees of feature congruence relate differently to audience engagement. The proposed framework extends the habituation–dishabituation theory, offers a practical approach to measuring content congruence in sponsored tourism videos, and provides useful guidance for content design and collaboration among tourism influencers, brand sponsors, and platforms.