When Service Fails, Can a Robot Win Back a Hotel Guest?

How would you feel, as a hotel guest, if you raised an issue and it was dealt with by a robotic worker? This increasingly realistic scenario is the setting for recent research by the SHTM’s Professor Janelle Chan and Professor YooHee Hwang. They provide the first evidence of how the growing use of robots in hospitality roles interacts with issues of timeliness and compensation when recovering from service failure. 

The involvement of robot workers in the service sector, driven by technological progress, cost pressures and chronic staffing gaps, has grown apace since the COVID-19 pandemic. “Although business is bouncing back due to the relaxed precautionary measures”, the authors note, “the hospitality industry still struggles with a severe labour shortage”. Thus, hotel guests in many countries are becoming used to receiving robotic assistance throughout their stay, not only at check-in. 

Indeed, with the integration of AI software into physical hardware, service robots are becoming better able to mimic human behaviour. This turn towards anthropomorphism could transform the customer experience. “For example”, the researchers note, “the Mandarin Oriental in Las Vegas has adopted Pepper, a humanoid robot providing interpersonal interactions”. Automated workers are no longer mere gimmicks: as anthropomorphic robots with real-time communication skills take on customer-facing duties, their acceptance by guests will be key to success. 

A rational, evidence-based approach to incorporating robots into hospitality roles is therefore essential. The authors were particularly interested in how guests would feel about humanoid robots in sensitive situations such as the aftermath of service failure. Their thinking was informed by the recently developed service robot acceptance model. This theoretical framework, they explain, “suggests that customers’ acceptance of robots is determined by both functionality and social-emotional elements”. 

A key question is how customers complaining of poor service react emotionally to robot versus human staff. Here, the researchers were further guided by another theoretical lens. “Social exchange theory”, they explain, “rests on the assumption that individuals engaging in an interaction believe that they can minimise costs and maximise benefits”. This has specific relevance to recovery from quality lapses. “In the service failure scenario, when individuals interact with service personnel, they are appraising the ratio of what has been given (i.e. loss) to what has been received (i.e. benefit)”. 

Human errors and system overwhelm are inevitable in busy service environments. Understanding the emotional dynamics of recovery – putting the mistake right to restore customer satisfaction – is thus essential for hotels. “When customers experience a service failure, they are motivated to rebalance an erratic emotion by receiving compensation”, the researchers explain. “Once customers are relieved of their negative emotion, the psychological consequence of forgiveness emerges”. 

However, regaining guests’ confidence may depend on more than simply offering compensation. “Customer forgiveness is not just about a comparison of cost and benefits but also its mediating relationship to trust”, the authors point out. Does compensation necessarily restore trust, and does it matter if the “employee” making the offer is a robot? Moreover, if a hotel’s response to service failure is delayed due to busy circumstances, does this affect the usefulness of compensation? Seeking answers, the researchers designed a pair of scenario-based experiments. 

Over 500 volunteers were enrolled in the tests. “Participants were asked to watch a video clip about a hypothetical hotel check-in experience”, the researchers report, “and imagine themselves as a customer in the video clip. Either a human or robot front desk agent helped them check in”. However, the “guests” subsequently discovered that their room was unclean. Upon complaining to the human or robotic agent about the room, some participants in this imagined scenario were offered, by way of compensation, a late checkout. Others were not afforded such compensation. 

The second experiment then “investigated the timeliness of service recovery as a boundary condition for the interaction between compensation and employee type”. Some participants had their rooms cleaned immediately, whereas the rest had to wait. This scenario was designed to explore whether a delay in rectifying the problem would alter the emotional dynamics of human- or robot-delivered compensation (or lack of compensation). In each case, the participants were subsequently asked to rate their forgiveness of the hotel and their trust in its service quality. 

The results of the first experiment showed that when the front-desk employee was human, customer forgiveness was higher when compensation was offered. Although this might not sound surprising in itself, it is only half the story: when the front-desk employee was a robot, compensation did not improve customer forgiveness. Evidently, financial recompense only has relational meaning when offered by a human. Moreover, the participant feedback regarding trust showed that “the mediating effect of forgiveness on trust was significant for human employees but not for robot employees.” 

What about timeliness? The results of the second experiment, in which the dirty room was cleaned either immediately or after a long wait, showed that “when delay exists in service recovery, customer forgiveness is higher after receiving compensation (vs. no compensation) from human employees, but not from robot employees”. In contrast, for those whose complaint was resolved immediately, the use of robots made no odds: “When there is no delay in service recovery”, the researchers explain, “customer forgiveness is not different between no-compensation and compensation conditions for robot employees [or] human employees”. 

“Timeliness in rectifying service failure”, they thus conclude, “is a more crucial factor than the humanlike appearance of a robot”. Their findings show that customers value the immediate resolution of problems as more crucial than either monetary compensation or human involvement in the recovery process. When offered by an automated worker, “compensation may not be effective in restoring the trust of robot (vs. human) employees”. 

Indeed, the situation is complex when a service lapse is compounded by a failure to solve the problem quickly. The authors refer to such a scenario – represented by the delay in waiting for the room to be cleaned – as “double deviation”. In such cases, monetary compensation can be effective, but only if offered by human employees. “Customers’ appraisement”, the authors surmise, “is not as rationalistic as social exchange theory presumes”. In other words, guests do not merely weigh up gains and losses but distinguish emotionally between human and robot interactions. 

The results have critical practical implications for hotels. Managers should consider whether it is wise to deploy robot employees when a timely response to service failure is unlikely, given their finding that “monetary compensation is effective in double deviation by human employees only”. At any rate, they argue, “businesses that deploy robot employees should minimise the use of compensation in retaining customer trust”. 

The acceptance of robotic staff in delicate communicative scenarios is evidently a nuanced issue. Harmonious workflows must be built in from the ground up. The researchers advise that “operational managers should work closely with robotic engineers in designing service robots with features to match their role in a specific service setting”. This requires service firms to focus on assessing and managing their customers’ expectations. Their study provides some of the first clues as to how this can be done. 

Janelle Chan and YooHee Hwang (2025). Trust and Forgiveness in Service: Effects of Single and Double Deviations with Human and Robot Staff. Asia Pacific Journal of Tourism Research, Vol. 30, No. 10, pp. 1401–1414.