Emojis as catalysts: How AI agents leverage emojis to enhance customer engagement across the pre-core and core service stages

Emojis, crucial nonverbal emotional cues, have gained scholarly attention in AI service marketing. However, existing research lacks a holistic investigation into the impact of emoji usage by AI from cross-stage perspectives. This study employs Emotions as Social Information (EASI) theory as the theoretical framework. We conducted three online experiments set in an OTA (Online Travel Agency) context. Study 1 (n = 264) adopted a 2 (Emoji usage by AI: yes vs. no) × 2 (Service stage: pre-core vs. core) between-subjects design. The results revealed that the use of emojis by AI service agents attracted customers' interest during the pre-core stage and enhanced customers' continuous usage intention in the core stage. Study 2A (n = 133) and Study 2B (n = 136) separate the pre-core stage and core stage to validate the mediating pathways of emoji effects. The results revealed that emoji usage by AI primarily functions through the affective pathway in the pre-core stage (i.e., evoking customers' pleasure emotions). In contrast, in the core stage, it primarily functions through the inferential process (i.e., triggering customers' social presence inferences about AI). These insights advance research on the benefits of emojis in AI services, offering implications for optimizing human-AI interaction experiences during the customer journey.