Sequential framework for analyzing mobile click-through decision in online travel agency with user digital footprints

In the hotel booking market, high click-through rates are essential for online travel agencies (OTAs) to earn commissions. Given the dominance of mobile devices in web traffic, analyzing the mobile click-through decision-making process plays a vital role in search engine optimization. This study proposes a sequential framework that leverages Bayesian inference to model individual users' click-through behaviors using user digital footprints, which encompass sequences of search, browse, compare, and click-through actions. This framework extracts three categories of information based on the degrees of dynamism in the hotel search process, ranging from less dynamic to highly dynamic levels: static hotel attributes, information cues in the search results, and temporal characteristics of user behaviors. Extensive experiments on a global OTA mobile clickstream dataset with over 600,000 observations reveal the substantial superiority of the proposed framework over the baseline models like probit regression and Naive Bayes. Notably, temporal characteristics emerge as the most important category. Drawing on our model, we delve into the interpretability of these three information categories. Additionally, we compare their varying impacts across different devices. Beyond these findings, this study offers valuable managerial implications for mobile OTA search engine marketing and optimization.