Trust in robo-advisory services: A mixed-methods study

This study examines how trust is formed in robo-advisory services through a two-study mixed-methods design. Study 1 uses PLS-SEM with cIPMA to analyse the effects and optimisation priorities of trust antecedents, while Study 2 employs semi-structured interviews to deepen and validate the findings. Results show that autonomy and intelligence enhance algorithm interpretability, interactivity, and structural assurance, whereas anthropomorphism has no significant effect. Autonomy is a necessary condition for effective uncertainty reduction, and both interactivity and structural assurance are necessary for perceptions of competence and warmth. Algorithm interpretability functions as a prerequisite to these perceptions. Both warmth and competence increase trust, but only competence is a necessary prerequisite. The qualitative insights provide contextual depth, revealing mechanisms and boundary conditions often missed in quantitative work. By integrating human-like characteristics and uncertainty-reduction strategies into a unified framework, this study advances understanding of trust formation in emerging financial services and offers insights for strengthening digital financial inclusion.