关键词: 电商直播, 人工智能, 虚拟主播, 购买决策, 在线口碑

Abstract: Powered by artificial intelligence technologies, virtual streamers have been widely adopted in Chinese livestreaming e-commerce as a scalable alternative to human streamers, enabling round-the-clock product promotion and real-time consumer interactions. Despite their growing industrial prevalence, the conversion effectiveness of virtual streamer livestreams remains highly inconsistent, and the mechanisms through which their interaction characteristics shape consumer behavior lack systematic academic investigation. This study examines how the interaction characteristics of virtual streamers influence consumers' pre-purchase decision-making and post-purchase word-of-mouth, aiming to reveal the underlying logic of converting livestream viewership into actual sales.
Grounded in an integrated analytical framework of informational and affective pathways, this study addresses three core research questions via empirical designs. First, it investigates the impact of virtual streamer visual attributes on consumers' pre-purchase behaviors, as well as the moderating effect of product information presentation modes. This part unpacks the cognitive processing mechanism through which visual design and information delivery jointly shape interaction quality and purchase intention. Second, the study explores how virtual streamers' communication styles drive pre-purchase consumer engagement through the affective interaction path, and identifies boundary conditions defined by consumer heterogeneity and product category. Third, it extends the analysis to the post-purchase stage by comparing consumer word-of-mouth evaluations of virtual and human streamers across varying service quality scenarios, and elucidates the asymmetric evaluation patterns and their psychological underpinnings.
This study makes three key theoretical contributions to the fields of livestreaming e-commerce and AI consumer behavior. First, from the interaction characteristic perspective, it structurally decomposes virtual livestreaming interaction cues into informational and affective dimensions, establishing a unified analytical paradigm that clarifies the differentiated roles of distinct interaction signals. This framework fills a critical research gap, as existing literature has largely focused on anthropomorphic avatar traits without systematically examining the core interaction process. Second, the study extends the analytical scope beyond pre-purchase intention to post-purchase word-of-mouth formation, revealing the asymmetric nature of consumer evaluations of virtual streamers under different service quality levels. This finding offers a novel theoretical account of the long-term relationship between AI agents and consumers, complementing the dominant focus on pre-purchase conversion in prior research. Third, it identifies heterogeneous boundary effects from both stable trait and situational state dimensions. Specifically, it introduces social overload as a dynamic situational variable to explore the adaptive boundaries of virtual streamer interaction strategies, addressing the limitation of existing studies that have paid insufficient attention to dynamic consumer states.
Practically, this study provides actionable implications for merchants, e-commerce platforms and digital marketing practitioners. For brand operators, the dual-pathway framework guides the optimization of virtual streamer design, including visual image customization and communication style matching, to enhance interaction efficiency and sales conversion. For livestreaming platforms, the findings support the development of virtual streamer configuration tools and post-purchase review management systems. For consumers, clarifying the interaction mechanisms helps mitigate irrational consumption caused by information asymmetry or affective persuasion, and improves overall purchase decision quality and post-purchase experience.
In sum, this study expands the theoretical boundaries of livestreaming e-commerce and consumer online behavior research, and provides theoretical support for the high-quality development of the digital economy and e-commerce sector. By integrating cognitive and affective mechanisms and covering the full consumer decision journey, it offers a comprehensive analytical framework for virtual streamer marketing and lays a solid foundation for future research on AI-enabled digital marketing interactions.

Key words: livestreaming e-commerce, artificial intelligence, virtual streamers, purchase decision-making, electronic word-of-mouth