Key points are not available for this paper at this time.
This study investigates how chatbot anthropomorphism influences customer purchase behavior through a real-world field experiment on a Japanese cosmetic e-commerce website. Using a 2 × 2 framework of form realism (human-like/cartoon appearance) and behavioral realism (warmth/competence response styles), the experiment examines how chatbot anthropomorphic designs influence one-time and subscription purchases over the short and long term. The findings show that low form realism (cartoon appearance) combined with high behavioral realism (warmth response) significantly enhances subscription purchases, aligning with the expectancy-disconfirmation theory. In contrast, chatbots with high behavioral realism (competence response) are more effective for one-time order customers, reflecting that cognitive cues are preferred in the exploration phase of customer–seller relationships. The study also reveals contrasting temporal patterns: while the novelty effect of chatbots diminishes over time for one-time purchases, subscription purchases exhibit no comparable decline, suggesting a persistent impact of chatbots in the maturity phase. These findings enhance the external validity of prior laboratory research and clarify the conditions under which low form realism can outperform human-like designs. The study demonstrates that consumers’ responses to chatbots evolve across repeated transactions, highlighting the importance of considering relationship phases and timeframes. No previous study has simultaneously examined both dimensions of realism; this study addresses this gap, offering theoretical insights into the effects of chatbot anthropomorphism and practical guidance for tailoring chatbot appearance and communication style to different customer segments.
Igarashi et al. (Tue,) studied this question.