Empathic chatbot for complaint handling in customer service

This research investigates the impact of empathy in chatbots on customers' perceived interactional justice during service recovery. Interactional justice refers to the perceived fairness of inter-personal behaviour, influenced by aspects such as the level of empathy of an employee, during service recovery. To this end, a complaint handling chatbot was developed based upon a combination of existing models of complaint handling and empathy for robots. A mixed-methods between-participant study was conducted with 25 participants, divided into two conditions: one interacting with an empathic chatbot (n=13) and the other with a non-empathic chatbot (n=12). The empathic chatbot includes several different empathic strategies, including use of empathic language, empathic intent mapping, emotion validation, and empathic questions. Using sentiment analysis to extract the user's dominant emotion from text, the chatbot is able to map its response to an appropriate empathic intent. It can also provide personalized responses based upon knowledge obtained through entity extraction from the conversation, increasing the overall feeling of empathy. The overall flow of the complaint conversation was designed using a decision tree, and a custom intent classifier was incorporated to optimize the chatbot's functionality. The findings showed that customers interacting with an empathic chatbot reported higher levels of interactional justice compared to those interacting with the non-empathic chatbot. Participants also expressed feeling more understood and helped by the empathic chatbot. Additionally, participants reported a more positive overall perception of the chatbot in the empathic condition, highlighting the possible positive effect of empathy on user experience. This research contributes to the understanding of the impact of empathy in chatbots for successful service recovery. By implementing empathic strategies, companies can enhance customer's perceived interactional justice, leading to increased customer satisfaction, loyalty, and overall user experience. However, this research also has its limitations. The study sample size is relatively small, which may impact the generalizability of the findings. In the future, studies may repeat the research with a larger sample size in an offline setting, explore the relationship between empathy and user engagement, and investigate the direct link between chatbot empathy and emotion regulation. In conclusion, this study has revealed that empathy in chatbots positively influences customers' perceived interactional justice.