Actions Speak Louder Than Words: Behavioral Versus Textual Credibility Signals in Tourism Review Platforms

Online reviews are central to traveler decision-making, yet their credibility is increasingly challenged by manipulated and strategically generated content. This study compares textual signals, or what reviewers say, with behavioral signals, or how reviewers act over time and across businesses. Drawing on more than five million Yelp reviews, the findings show that behavioral patterns, especially temporal activity and reviewer-business structure, provide more reliable indicators of manipulation likelihood than textual features alone. A late-fusion model combining behavioral and textual evidence offers the strongest overall assessment, while category-level results show that manipulation risk is unevenly distributed across tourism contexts. The study advances tourism research by framing review credibility as a multi-signal assessment problem and by highlighting behavioral consistency as a foundation of trust. Practically, platforms should prioritize behavioral monitoring, review-velocity tracking, and reviewer-business pattern analysis while using text-based detection as a complementary governance layer.