Voice Shopping: The Effect of the Consumer-Voice Assistant Parasocial Relationship on the Consumer's Perception and Decision Making

Abstract

The primary goal of this study was to investigate how shopping with voice assistants may be uniquely different from shopping on websites. This study focused on whether using different shopping mediums (i.e., voice assistant and websites) affects the way consumers evaluate the recommended product offered by the shopping medium. Based on the anthropomorphism literature and the parasocial interaction theory, the study proposed consumers to form a stronger parasocial relationship with a more humanlike shopping medium, which in turn influences consumers to evaluate the recommended product more positively. Specifically, consumers were expected to perceive voice assistants as more humanlike than websites because of the way voice assistants are designed (i.e., vocal conversation). Furthermore, the study aimed to understand the effect of two moderators, interaction style (task-oriented interaction vs. socially-oriented interaction) and product type (search product vs. experience product). To investigate the following questions, two experimental studies were conducted. Both studies recruited participants who are 18-36 years old and are familiar using voice assistants. Study 1 (N=85) utilized a 2 (shopping medium type: voice assistant vs. website) x 2 (interaction style: task-oriented vs. socially oriented) between-subject experiment factorial design. Participants were invited to the lab to interact with Amazon Echo or the Amazon website. Their interaction styles were manipulated using instructions that are focused on either socially-oriented interaction or task-oriented interaction. Study 2 (N=418) utilized a 2 (shopping medium type: voice assistant vs. website) x 2 (product type: experience product vs. search product) between-subject online experiment factorial design. Study 2 participants were recruited via Amazon MTurk. In Study 2, a hypothetical retailer was created instead of using currently available voice assistants and websites to eliminate the effect of preexisting relationships on the results. The recommended products were manipulated by two products with different search qualities and experience qualities. In both studies, the results of MANCOVA/MANOVA and PROCESS mediation analyses revealed that consumers evaluated products more positively when they were recommended by the shopping medium they formed a stronger parasocial relationship with. Consumers developed a stronger parasocial relationship with the shopping medium they perceived to be more humanlike. However, unlike hypothesized expectations, consumers perceived websites to be more humanlike than voice assistants, consecutively formed a stronger parasocial relationship with websites and evaluated products recommended by the websites more positively. The moderating effect of interaction style was not statistically significant, but the moderating effect of product type was statistically significant. Participants in the website condition evaluated the recommended experience product significantly more positively than participants in the voice assistant condition. Their evaluation of the recommended search product did not vary significantly between the website condition and the voice assistant condition. The findings suggest people may perceive voice assistants as an autonomous agent apart from their operating brands while perceiving websites to be inseparable from their operating brands (e.g., employee, product, CEO). In addition, although the proposed hypotheses were not supported, the findings still support the proposed model that suggested consumers be persuaded more by the more humanlike shopping medium because they form a stronger parasocial relationship with it. Further, the findings also suggest a recommended product’s search or experience qualities may critically influence the way consumers evaluate it. The research contributes to the anthropomorphism literature and parasocial interaction theory by confirming the causal relationship between humanlikeness and parasocial relationships. Further, the research provides knowledge related to utilizing voice assistants in the field of consumer behavior.