Fleeting or Lasting Success? Unlocking the Power Durability of Live Commerce

Abstract

Live streaming has emerged as a powerful marketing tool for product promotion, with increasing adoption and usage frequency among e-commerce sellers. However, the heavy reliance on live streaming raises questions about its cost-effectiveness and the durability of each live streaming session's impact. This study answers these questions by examining the temporal patterns of individual live streaming session's sales effects and unraveling the mechanism (i.e., the distinctive nature of live commerce) that underpins such patterns. Collaborating with a large Asian live commerce platform, we synthesize a quasi-natural experimental setting and reveal a significantly positive yet short-lived sales effect, with no lasting effects beyond the session day unless followed by another live streaming session. This pattern indicates that real-time interactions, which bring entertainment value and engagement, play the primary and distinctive role in driving sales and generating value within frequent live streaming contexts. This mechanism overshadows potential cascade effects from increased awareness through influencers or learning effects due to reduced uncertainty through video content. Further analysis regarding the necessary conditions for these mechanisms indicates that the sales boost is moderated by the level of engagement during sessions, rather than the extant information about the seller and product. Also, the first session does not exhibit an incremental sales effect. Together, these findings reinforce our proposed mechanism. These findings add to the literature on live commerce and influencer marketing, shedding light on the evolution of live commerce. They also provide valuable guidance for the strategic use of live streaming, including considerations of timing, format, influencer selection, and interplays with other marketing strategies, which is crucial for live streamers, marketers, sellers, and regulators.

Keywords: Live Streaming, E-commerce, Quasi-experiment, Temporal Pattern