Purpose – Despite having been widely studied in traditional (bricks-and-mortar) services, the effect of service failures and recovery (SFR) on customer loyalty has received only limited attention in the context of e-services. This paper sets out to empirically test the following set of hypotheses in an e-service setting: H1, service failures have a negative effect on customer loyalty intentions; H2, failure resolution has a positive effect on customer loyalty intentions; H3, satisfaction with the recovery has a positive effect on customer loyalty intentions; H4, outstanding recovery results in loyalty intentions which are more favorable than they would be had no failure occurred (service recovery paradox). Design/methodology/approach – The paper is based on an online survey of actual customers of a commercial e-banking service. Findings – H1-H3 are supported, suggesting that: the detrimental effects of failures are also present online; problem resolution leads to increased loyalty; despite the challenging nature of online failures and the reduced degree of human interaction, it is possible to achieve effective recovery in e-services. H4 is also supported. We observes a recovery paradox effect but it only take place for a small proportion of “delighted” customers, i.e. those who perceived an outstanding recovery. Although unlikely, the impact (size effect) of outstanding recovery on loyalty is substantial. Research limitations/implications – Future research should examine other types of e-services. Practical implications – E-service delivery systems should be designed with a strong failure-prevention mindset and include effective service recovery mechanisms. However, in general, e-service providers should not look at superior recovery as a substitute for error-free service. Despite not being a viable strategy in general, delighting customers in the recovery may make sense for the most profitable customers. Originality/value – The paper provides empirical evidence of the effects of SFRin the context of online service, an area which has received limited attention to date. Unlike other research, this paper draws on data from customers of an actual e-service and therefore benefits from increased external validity
Stored only when both papers are in the corpus — 128 citations exist upstream.