As if human or as if machine? Examining dimensional trade-offs of machine heuristics in human-virtual influencer collaborations

Purpose Virtual influencers (VIs) increasingly collaborate with humans or other VIs in branded content, yet how such collaborations shape consumer responses remains unclear. This research examines which machine heuristic dimensions are activated by collaborator type and contextual cues, and how these dimensions mediate persuasion outcomes. Design/methodology/approach Four online experiments systematically manipulated collaborator type (human vs. virtual), relational closeness, and gender congruence. Across studies, machine heuristic dimensions and consumer responses, including attitudes toward posts and intentions to follow the VI, were measured. Findings VI collaborations do not uniformly enhance humanization. Instead, they differentially activate specific machine heuristic dimensions. Human collaborators reduced perceived unemotionality and, under certain conditions, increased perceived accuracy, which in turn predicted more favorable responses. Relational closeness mitigated unemotionality, and gender congruence shaped whether human collaborators amplified or weakened accuracy perceptions. Collaboration outcomes thus depend on which heuristic dimensions are activated. Originality/value By integrating CASA with a multidimensional account of machine heuristics, this research shows that collaboration cues generate dimensional trade-offs rather than uniform effects. The findings offer a more precise theoretical account of VI persuasion and actionable insights for collaboration design in interactive marketing.