A big data and machine learning-based approach to theorising consumers’ continuance intentions to use digital wallets

This study examines factors associated with user engagement and continuance intention towards Saudi digital wallet applications. We adopted a sequential mixed-methods approach to analyse 23,830 cleaned user-generated reviews from Google Play and the App Store. We used Latent Dirichlet Allocation (LDA) and sentiment analysis to extract key themes, which we integrated with a structured literature review and expert-supported construct mapping to develop a conceptual framework grounded in the IS Success Model and complemented by a multidimensional perspective on user engagement. This framework was empirically evaluated using Lasso- and Ridge-regularised regression techniques. Findings indicate that service quality, trust, information quality, and technical support effectiveness are strongly associated with user engagement, which, in turn, is positively associated with continuance intention. The study extends the IS Success Model by incorporating trust and technical support effectiveness as contextually relevant post-adoption factors and by positioning engagement as a distinct cognitive, emotional, and behavioural mechanism rather than satisfaction or habitual use. Methodologically, it illustrates the value of integrating NLP, theoretical interpretation, expert validation, and regularised regression to examine theoretically specified relationships using naturally occurring user feedback.