A natural apology is sincere: Understanding chatbots' performance in symbolic recovery

Although chatbots have been widely used in dealing with service complaints, knowledge about the recovery performance of chatbots is limited. Drawing upon lay belief and emotional competence theory, this research explores symbolic recovery performances of chatbots in two experimental studies. The results show that symbolic recovery from chatbots leads to lower customer satisfaction than symbolic recovery from human employees due to the lay belief that chatbots lack emotional competence. Perceived naturalness and perceived sincerity play a sequential mediating role. Customers perceive chatbots' symbolic recovery to be less natural than that of human employees. Less natural recovery is perceived to be less sincere, thus decreasing customer satisfaction with the recovery. Changing perceived diagnosticity of the lay belief can improve customer satisfaction with chatbots' symbolic recovery. This research enriches theoretical research on symbolic recovery and chatbots, providing information for how companies can effectively use chatbots to make an appropriate recovery. • The study explores the recovery performance of chatbots. • Using chatbots to handle customer complaints leads to lower satisfaction relative to humans usually. • Chatbots' apology is viewed as unnatural and insincere by most customers. • Changing the beliefs that chatbots lack emotional competence can improve customer satisfaction in this area.