Abstract
Telecom mobile apps have become one of the core channels through which telecom operators provide value-added services to customers as a strategic revenue stream. Therefore, understanding consumer feedback through online reviews for these service channels is imperative for developing sustainable strategies to retain consumers. Helpful online reviews are instrumental in shaping consumer decision-making and have become a core research theme across domains, with a growing interest in determining what makes a review helpful to consumers. This study sought to identify the determinants of helpful reviews in the context of telecom mobile apps. Using a combination of text mining and machine learning, this study identified how review characteristics and review content influence the helpfulness of telecom mobile apps. In particular, review length, review age and review response are among the most influential determinants of helpfulness. The study also illustrates how the seven topics identified with Latent Dirichlet Allocation (LDA) influence review helpfulness. Additionally, SHapley Additive exPlanations (SHAP) interaction values illustrate how the content of a review explains the link between review length and helpfulness. Longer reviews that focus on performance issues and transaction experience were seen to be more helpful, whereas those that focus on app quality and user satisfaction were less helpful. The study concludes by exploring the implications for theory and practice.