Service recovery strategy in platform economy: a moderated-moderation model with compensation, punishment and aggregator’s effort

Purpose Service failure and recovery in platform economy is under-researched. This study aims to investigate how aggregators’ recovery actions impact customer satisfaction, negative word-of-mouth (nWOM) and repurchase disinclination when service failures are caused by actual service providers (ASPs). Design/methodology/approach Using three studies (one retrospective sampling-based survey, one experiment and one post hoc qualitative interview) based on a total sample size of 596 respondents we validated our proposed model. Findings Compensation improves satisfaction but weakens with punitive actions. Aggregator efforts moderate these effects, reducing nWOM. Complaint motivations differ by issue severity, influencing satisfaction, repurchase disinclination and platform perceptions in unique ways. Research limitations/implications This study advances service recovery research by highlighting triadic responsibility attribution, the ambivalent role of punitive actions and cognitive reframing through aggregator efforts. It extends justice theory in platform ecosystems, emphasizes severity-contingent complaint outcomes and challenges the satisfaction–repurchase link, offering a comprehensive justice-based framework for aggregator-mediated service failures. Practical implications Aggregators must balance compensation with visible yet empathetic punitive actions, especially in severe service failures. Transparent recovery efforts can curb nWOM even if satisfaction declines. However, overcompensating or excessive interventions may erode trust and deter repurchase. Tailoring recovery strategies by issue severity is vital for sustaining customer relationships. Originality/value To the best of the authors’ knowledge, this is probably the only study that focuses on the comparative effectiveness of various recovery strategies in a platform economy context where ASP is responsible for service failure.