What makes customers “Click”? An analysis of hotel list content using deep learning

Abstract

Understanding what motivates customers to click through a hotel on the search results page for more details under the influence of COVID-19 is important since it is a prerequisite step before booking a hotel. This study aims to examine how visual and textual information cues shown on each hotel listing influence customers’ click-through behavior using real-time viewing times as an indicator. Using the deep learning DenseNet model, this study analyzes both visual and textual information cues from 3617 hotel listings and reveals that: 1) Four hotel cover photograph features (building outside view, pool view, street view, property detail) extracted significantly influence click-through behavior except for pool view; 2) Presenting social proof badges (guest favorite, very clean) and booking terms and conditions (pay later, free cancellation) are effective marketing tactics. This study enriches the literature on hotel information search behavior and assists hotel practitioners to optimize their listing information.