Artificial Intelligence Recommendations and Booking Abandonment in Online Hotel Reservation Platforms: A Cognitive Load and Information Foraging Perspective

I-driven recommendation systems are widely assumed to ease decision-making in online hotel booking, yet little is known about whether two of their defining characteristics—perceived serendipity (unexpected but relevant suggestions) and perceived similarity (resemblance among suggested properties)—impose cognitive costs that contribute to booking cart abandonment. Objective: This study examines how perceived serendipity and perceived similarity in AI recommendations influence cognitive load, and how this load in turn affects booking abandonment, drawing on Cognitive Load Theoryand Information Foraging Theory. Methodology: Cross-sectional survey data were collected from 624 Indian consumers with recent AI-assisted hotel booking experience, sampled from the user communities of three online travel platforms, and analyzed using partial least squares structural equation modeling (PLS-SEM) in SmartPLS 4.0. Results: AI-driven personalization increased both perceived serendipity (β = 0.777) and perceived similarity (β = 0.824), each of which independently elevated cognitive load (β = 0.518 and β = 0.464, respectively); cognitive load in turn was significantly associated with greater booking abandonment (β = 0.530), and the indirect effects of serendipity and similarity on abandonment via cognitive load were also significant. Conclusions: In this exploratory, single-sample study, AI personalization did not uniformly ease booking decisions: when recommendations were perceived as highly serendipitous or excessively similar, the associated cognitive load coincided with greater cart abandonment. These preliminary, cross-sectional associations tentatively suggest that hospitality platforms may benefit from calibrating personalization intensity rather than maximizing it, pending confirmatory replication with stronger measurement-invariance and discriminant-validity testing.