Introduction and Research Question
Even though there is increased interest in service robots from both practitioners and scholars (Wirtz et al., 2018; Lu et al., 2020), the knowledge concerning how robots influence consumer behavior in retail settings is still scarce and thus much needed (Shankar, 2018; Biswas, 2019; Lu et al., 2020; Belanche et al., 2021). In particular, there is only little research on the impact of sensory information during the interaction with a service robot (Biswas, 2019). Haptic factors like touch have mostly been overlooked in research on human-robot interaction (Willemse et al., 2017; Law et al., 2021; Hayashi et al., 2022) despite their crucial role for human-robot bonding and communication (Andreasson et al., 2018) and customer experience (Lemon and Verhoef, 2016). Interaction between a service robot and the customer can take place through the modality of touch, in that not only customers touch robots but also robots touch customers (Law et al., 2021). Due to the importance of touch in everyday social interactions (Gallace and Spence, 2010), it is essential for retailers to understand how customers would react to being touched by a service robot. Trust has been considered one of the core responses when studying robots in socials contexts (Law et al., 2021) and is considered to influence customer experience (Lemon and Verhoef, 2016). Different types of touch might have different meanings for individuals and are thus evaluated differently by the customer. Moreover, robots, in general, might take over tasks during a human-robot interaction (De Gauquier et al., 2020). For instance, they might provide customers with information, guide them through the store, point them to the location of specific products and recommend products (Barnett et al., 2014). How customers respond to different touch types and shopping assistance by a service robot has not been answered yet in retail and service research.
We put forth hypotheses as to the effects of (1) touch type, (2) assistance by a service robot and (3) interaction between touch type and assistance by a service robot on consumers’ trust in the service robot. Moreover, we hypothesize that the effect of touch type can be explained by perceived interaction comfort.
A usable sample of 245 German consumers (mean age = 28.38 years; 31 % male) was recruited to take part in the study. We employed a 3 (robot-initiated touch type: hug, handshake, no touch/waving) x 2 (assistance from the robot: yes, no) between-subjects factorial design where participants were randomly assigned to one of the six scenarios. Each participant received a scenario and a questionnaire. The scenario described a shopping situation and participants were asked to put themselves into the described situation. The humanoid robot Pepper developed by Softbank Robotics was chosen to portray the service robot in the present study (SoftBank Robotics, 2020). The questionnaire comprised realism checks, manipulation checks, perceived trust in the robot, perceived comfort as well was potential covariates and a standard set of socio-demographic questions. Wherever possible, we used seven-point Likert scales anchored by 1 (“strongly disagree”) and 7 (“strongly agree”). To test our hypotheses, we conducted an ANCOVA with trust in the robot as the dependent variable and the manipulations of touch types and assistance together with their interaction as independent variables. Based on prior research, we included gender (Stier and Hall, 1984) and technology readiness (Parasuraman, 2000) as covariates. We further tested whether the effect of touch type on trust was mediated by perceived interaction comfort. Mediation was performed according to Hayes (2013).
Results indicate that touch types that violate social norms and/or a customer’s expectation as to what is appropriate in the shopping context leads to a decrease in trust in a humanoid service robot, which can be explained with the amount of (dis-)comfort felt. Moreover, participants trust the robot more, when the robot provides shopping assistance compared to no assistance. Further, providing assistance to the customer by the robot may change the effect of different touch types on trust in the robot.
First, it is essential for service providers and retailers to understand how customers would react to being touched by a service robot and how and why they would react to different touch types. Different types of touch might have different meanings for individuals and are thus evaluated differently by the customer. We provide evidence of the associated underlying process. Second, robots might take over certain tasks during a human-robot interaction (De Gauquier et al., 2020) and thus offer assistance to the customers. This might lead to different evaluations as well. Third, we examine if different combinations of touch types and assistance may affect customer evaluations differently. Our findings are of importance to retailers and other service providers that want to know how robot-initiated touch and assistance can enhance customer interactions at the point of sale.
The findings of this study contribute to the emerging research field of human-robot interaction, with focus on human-robot touch, and provide important insights on how to employ service robots in retail stores, especially regarding physical interaction with the customer.
Research limitations and outlook
We examined the effects of robot-initiated touch types and assistance based on scenario descriptions and pictures. The participants of the study had to imagine the described interaction with the service robot. A better option would be that participants actually experience the interaction with the service robot and really feel the touch from the robot. Moreover, the study only included service robots as service agents. Although this procedure was useful in order to find out how to employ service robots, it does not reveal the differences compared to human employees in the store. Further research should also examine how the effects of robot touch differ from those caused by human touch in the retail setting