From Voice to Choice: Understanding the Consumer Response to Voice Assistant and Human Recommendations

Abstract

Information technologies, particularly voice assistants (VAs), have transformed consumer interactions by enabling voice-activated tasks and product recommendations. However, existing research has not fully explained the mechanisms underlying consumer responses to VA versus human-generated recommendations. This research addresses this gap by investigating whether VAs influence consumer intentions differently from human salespeople. Using a randomized between-subjects experimental design, we conducted two single-factor scenario-based studies (n = 614), in which participants received a recommendation either from a salesperson or a VA. Results reveal that although consumers report greater willingness to use VAs in the future, recommendations from human salespeople more strongly influence both the intention to follow the recommendation and to purchase the product. Perceived credibility and warmth emerged as key explanatory mechanisms, with human salespeople rated higher on both dimensions. Moreover, we find that need for cognition moderates these effects: individuals high in need for cognition respond more critically to VA recommendations than those low in need for cognition. These findings contribute to a deeper understanding of how consumer perceptions and individual traits shape responses to AI versus human interactions, while offering actionable insights for improving VA design to increase warmth and credibility.