Predicting key touchpoints in hotel customer journey–a comparison of machine learning models

Abstract

This paper investigates machine learning’s role in predicting key hotel touchpoint interactions across their journey, improving customer lifetime value and loyalty. Prior studies focused on cancellations and revenue, neglecting other guest interactions. Using data from a resort hotel and a city hotel, we employ several algorithms, achieving recall scores over 80% for cancellations, F1 Scores of 66% and 85% for food package predictions, and AUC and recall rates exceeding 90% for rebooking. Variables such as lead time, deposit type, booking changes, and previous cancellations are fundamental for our models, contributing to the literature of predictive capabilities in hospitality.