AI-personalised recommendations and brand trust: a dual-appraisal account of perceived diagnosticity and algorithmic persuasion awareness

Purpose This study examines the joint associations of perceived AI-personalised recommendations, perceived diagnosticity and algorithmic persuasion awareness with trust in the recommended brand. Design/methodology/approach A cross-sectional survey of 412 consumers in China who recalled a recent recommendation they perceived as AI-personalised was analysed using partial least squares path modelling. Findings Perceived personalisation was positively associated with diagnosticity (β = 0.472) and algorithmic persuasion awareness (β = 0.353). Diagnosticity was positively associated with brand trust (β = 0.413), whereas algorithmic persuasion awareness was negatively associated with it (β = −0.194). The corresponding indirect associations were 0.195 and −0.069. Research limitations/implications The cross-sectional, self-reported design limits causal and temporal interpretation. Experiments, observed exposures and longitudinal studies can test the proposed relationships and boundary conditions. Practical implications The findings suggest potential applications in recommendation design involving decision support, transparency and consumer control alongside behavioural performance. Social implications Consumer education could help consumers distinguish product relevance from the data use and commercial purpose underlying personalised recommendations. Originality/value The study extends recommendation research by examining perceived diagnosticity and algorithmic persuasion awareness jointly in relation to trust in the recommended brand. This focus complements work on recommendation-agent trust, trust in ad information and AI-related information disclosure.