Cold Comfort Matters - How Channel-Wise Emotional Strategies Help in a Customer Service Chatbot

Smart service chatbots, aiming to provide efficient customer service, have increased rapidly. Currently, few task-oriented chatbots had employed emotion strategies together with problem solving process. Meanwhile, customers can have different preference across channels (e.g., social media and webpages). This paper presented the design for a series of emotional strategies (ES) combined with task-solving, and tested its effectiveness across channels. In the multi-channel service chatbot, Moli, we compared ES with brief apology (BA) and no emotion responses (NER) in a Wizard of Oz study. The results indicated the effectiveness of the emotion strategies borrowed from psychotherapy technologies. Meanwhile, we found customers preferred brief and direct strategies on webpage, whereas rich emotional strategies in all stages were more favorable on social media. With detailed strategy design, the "cold comfort" was found helpful for building human-machine relationships, pacifying emotions, and improving satisfaction and system usability.