Polite or playful? How service robot type and apology style influence negative word-of-mouth after service failures

As service robots become increasingly common in customer-facing roles, effective service recovery after failures is becoming crucial for maintaining customer trust. Drawing on expectancy violation theory, this research examines how robot type (warm vs. competent) and apology style (polite vs. humorous) jointly shape negative word-of-mouth (WOM) responses. Across three online scenario-based experiments, Study 1 demonstrates that apologies are most effective when apology style aligns with robot type: humorous apologies reduce negative WOM for warm robots, whereas polite apologies are more effective for competent robots. Study 2 identifies expectancy violation as a key psychological mechanism underlying these effects, showing that incongruent combinations amplify negative WOM through heightened expectancy violation. Study 3 further reveals a boundary condition of failure severity, such that congruent apologies are more effective in low-severity failures, while polite apologies dominate in high-severity contexts regardless of robot type. These findings advance expectancy violation theory by integrating robot typology and communication strategies into a unified explanatory framework and offer actionable guidance for designing context-sensitive service robot communication to manage customer expectations and mitigate negative WOM.