Recovering customer satisfaction after a chatbot service failure : The effect of gender

Titelangaben

Rese, Alexandra ; Witthohn, Lennart:
Recovering customer satisfaction after a chatbot service failure : The effect of gender.
In: Journal of Retailing and Consumer Services. Bd. 84 (2025) . - 104257.
ISSN 0969-6989
DOI der Verlagsversion: doi.10.1016/j.jretconser.2025.104257

Volltext

	Format: PDF Name: 1-s2.0-S0969698925000360-main.pdf Version: Veröffentlichte Version Verfügbar mit der Lizenz Creative Commons BY 4.0: Namensnennung
	Download (2MB)

Abstract

Chatbots in customer service often fail to meet customer expectations, largely because they are considered prone to comprehension errors. Service recovery can decisively restore perceived humanness and user satisfaction through perceived warmth and competence after a service failure. In this study, we investigate the effect of the chatbot's gender on the user in service recovery. The majority of chatbots in customer service display female characteristics. We use a pre-study (n = 30) to determine the perceived gender of several chatbot avatars and a scenario-based experiment (n = 300) in which the service recovery after an outcome failure and the gender of the chatbot are manipulated. The results show that the service recovery significantly improved user satisfaction with the chatbot. In addition, the chatbot was perceived as significantly warmer and more competent, which resulted in higher perceived humanness and increased user satisfaction. Male chatbots were perceived as less warm in failure situations when service recovery was not achieved. However, following service recovery, there are no differences in the perception of the chatbot's warmth and gender. Perceived warmth is correlated with perceived competence. Gender incongruence between the chatbot and the respondent resulted in a higher perceived humanness of the chatbot in service recovery. Therefore, firms should pay particular attention to the contexts in which chatbots are used and whether gender matching is appropriate.