E-marketplace live streaming (ELS) has emerged as a popular third-party service for improving sales performance. It often persuades consumers through streamers’ storytelling, that of descriptions and depictions of products and their uses, which encompass narrative messages describing the experience of streamers. Little is known, however, about how narration can improve the sales performance of streamers. Extending narrative persuasion theory to the ELS context, we posit that narrative content—temporal-causal and first-person embedding, as well as narrative delivery—product display and interaction response—can improve sales performance through their influence on consumers’ imagination of plotline or empathy for streamers. We also posit that the efficacy of narrative content elements is contingent respectively on narrative delivery elements. Using dictionaries, natural language processing, and computer vision techniques, we will extract these four variables from videos data obtained on Taobao Live platform. We will then empirically quantify their effects with model estimations.