Chatbots in service recovery: Crackpot or Jackpot?

Abstract

Problem Service recovery has been a heavily researched and studied area for several years. As technology has continued to develop and been adapted and incorporated into this field, new research avenues have also opened up, many of which have not been studied in much depth as of yet. With very limited research on firms using chatbots as a service encounter in service recovery, we wanted to investigate this further. Therefore, we wanted to check if there was any difference between a customer interacting with a human or a chatbot in a service recovery situation and measure this with regard to satisfaction with the firm and recovery. Purpose The purpose of this study is to analyse the effect of an unsuccessful and a successful service recovery, provided by either a human, chatbot or a combination of the two, on satisfaction with the firm and service recovery. Research A scenario-based survey experiment was chosen to answer the Design research problem and questions for this thesis. The design used was a 3 (Chatbot, Human, Chatbot+Human) x 2 (Unsuccessful, Successful) between subject design. Findings Customers are proven to be more satisfied when interacting with a chatbot in a successful service recovery, than interacting with a human or a combination of the two. However, when the customers interact with a chatbot in a unsuccessful recovery, they are even more unsatisfied than when interacting with a human or the combination. We also found a significant relationship between the outcome variable and satisfaction with recovery and firm, and a significant interaction effect between the outcome and encounter variables and satisfaction with firm and recovery. Keywords Service recovery, customer satisfaction, chatbot, human, service failure, E-commerce, anthropomorphism,