Engagement Depth and Booking Intent in AI-Mediated Tourism Discovery: Evidence from a Regional Destination Portal

Tourism’s digital transformation has reshaped how travelers search for and evaluate destinations. However, relatively little empirical work has examined how user engagement translates into booking intent, especially under the emergent discovery channels mediated by artificial intelligence (AI). This study tests an engagement-driven referral framework using longitudinal behavioral data from a Mediterranean destination portal (April 2022–January 2026; 1.6 million sessions). Engagement depth, measured as average session time, significantly predicts booking intent click rate. Mobile drives 83% of sessions, but desktop users convert at nearly twice the rate (5.69% vs. 3.37%). High traffic, as it turns out, does not equal high commercial intent. Lower-volume international markets routinely outperform the dominant domestic market. The most striking result concerns AI referrals. Traffic arriving from AI assistants converts at 8.26%, more than double the organic search rate of 3.88%, despite shorter sessions, a pattern consistent with compressed decision-making under generative AI. These findings, grounded in real travel portal data, extend engagement theory beyond transactional settings and shed early light on how referrals from AI assistants like ChatGPT or Gemini differ behaviorally from organic search, with practical implications for portal managers, destination marketing organizations (DMOs), and sustainable demand management.