This study presents an empirical research which main purpose is to determine the variables with drive loyalty of mobile banking users, to segment consumers and to explain the causes. A self-administered online survey was used to collect data later assessed through partial least squares analysis. The partial least squares-prediction-oriented segmentation methodology evidenced unobserved heterogeneity. Also, the most significant variables within each segment were revealed through an importance-performance map analysis. This general model shows that the current banking usage and satisfaction have a positive impact on customer loyalty. Three different segments were obtained according to banking usage, ease of use and utility variables. Satisfaction turned out critical in all cases. Research was conducted in only one country (Chile). Thus, limited in size but large enough for the PLS analysis. The conclusions and managerial implications provide alternatives and strategies for banking entities so they will need to approach satisfaction measurement and enhancement systems alongside discriminatory variables to increase trust, utility and ease of use in every segment. As of today, customer segmentation in mobile banking was based on a priori partitions. This new approach does not depend on sample distribution, it rather determines unobserved heterogeneity. In addition, this research also evaluates current mobile banking usage.