Do emotional reviews predict sales? Evidence from an AI-powered experimental investigation

Abstract

Abstract In the digital age, online reviews have emerged as a pivotal touchstone for consumer decision-making, with the emotional sentiment of these reviews playing a crucial role in shaping purchasing behaviors. This research harnesses signaling theory and the elaboration likelihood model to examine the influence of emotional sentiment in online reviews on consumers’ purchase behavior. Through the application of artificial intelligence via natural language processing, we analyzed 14,585 reviews from a paid knowledge platform, which, in turn, enabled us to construct a rich dataset of reviews with varied emotional sentiments for a between-subjects scenario-based experiment. This research reveals a pronounced influence of positive sentiment on purchase behavior, which significantly eclipses the effects of mixed or negative sentiments. Positive sentiment significantly increases the perceived credibility and quality of reviews relative to both mixed and negative sentiment, whereas mixed and negative sentiment do not significantly differ on these perceptions. Despite investigating the potential for serial mediation by the credibility, quality, and usefulness of reviews, the findings suggest a more direct pathway by which emotional sentiment influences purchase behavior, challenging preconceived notions about the complexity of this relationship. Theoretically, this research enriches the discourse on the role of emotional sentiment within the frameworks of signaling theory and the elaboration likelihood model, offering new insights into the drivers of consumers’ purchase behavior in the digital marketplace. Practically, the implications of this research provide actionable strategies for e-commerce practitioners and digital marketers to strategically manage online review ecosystems, leveraging the power of emotional sentiment to influence consumers’ purchase behavior positively.