This study aims to examine how intelligent service agent (ISA) appearance (human-like vs. machine-like), product type (utilitarian vs. social), and interaction frequency (high vs. low) jointly shape consumer responses to AI-generated recommendations.
Across three experiments and a focus group study, this research tests the interactive effect between ISA appearance and product type, identifies processing fluency as the underlying mechanism and examines interaction frequency as a key boundary condition. ANOVA, ANCOVA and moderated mediation analysis was conducted.
Results demonstrate that, machine-like ISAs are more effective for utilitarian products, whereas human-like ISAs are more effective for social products. This effect is explained by processing fluency and varies with interaction frequency. Finally, focus group interviews further support and contextualize the experimental findings.
This study focused on utilitarian and socially expressive product types, yet alternative categorizations (e.g. creativity or hedonicity) warrant investigation. Potential boundary conditions, such as consumers' maximizing tendencies or loneliness, were not examined and may moderate the observed effects. Methodologically, the use of lab experiments and focus groups may limit external validity. Future research should explore diverse product taxonomies, psychological moderators and real-world behavioral data from digital platforms to enhance the generalizability and contextual relevance of the findings.
This study offers actionable insights for firms utilizing ISAs in digital recommendation contexts. Firms should strategically align ISA design with product type – deploying machine-like ISAs for utilitarian products and emphasizing analytical competence, while using humanoid ISAs for socially expressive products to enhance emotional resonance and identity alignment. Matching ISA anthropomorphism to product characteristics can improve recommendation effectiveness, foster consumer trust and optimize service efficiency. Moreover, ISA adoption can reduce labor costs without sacrificing personalization, provided that firms tailor ISA cues to consumer expectations and product attributes.
As ISAs increasingly mediate digital interactions, understanding their social impact is critical. This study highlights how anthropomorphic design can influence consumers' psychological processing and engagement with AI, raising important considerations around trust, emotional connection and digital inclusion. Properly aligning ISA design with product type not only enhances marketing effectiveness but also fosters more human-centered and socially resonant AI interactions. These findings contribute to broader societal discussions on human–AI collaboration, suggesting that thoughtful ISA design can support more empathetic, accessible and socially attuned digital environments.
This research advances interactive marketing by moving beyond human–AI comparisons to theorize heterogeneity within AI agents, positioning ISA appearance as a central interactive cue and providing a process-based explanation of how AI design shapes consumer responses in dynamic brand interactions.
