Emotional vs Rational Robot Shopping Assistants

As digitalization advances and online shopping rises, traditional retail faces significant hurdles. While self-checkout kiosks have become commonplace, many consumers still prefer the personal touch and guidance they receive in store, especially for more expensive purchases such as electronics. Surveys indicate that a substantial portion of shoppers (around 54%) research smartphones in store before making a purchase decision. Historically, human sales associates handled consultations, but recent strides in robotics, AI, and advanced language models now enable retailers to deploy social robots capable of blending human-like interaction with machine efficiency. In countries like Japan, known for its robotics innovation, the use of robots (such as Pepper) as sales assistants is increasingly prevalent, with Western retailers also dabbling in robotic experimentation. For instance, the German electronics chain Media Markt has introduced robots in select stores to provide product information and guide shoppers to specific items. This trend raises the question of how consumers respond to robot-based sales assistants and how such interactions influence their purchasing decisions.

This research aims to examine how the design of robot-based sales assistants impacts consumer trust and purchasing behavior in retail settings. We are particularly interested in contrasting two design approaches: emotional versus rational. For this purpose, we developed a robot named “Saleshat” using the Furhat robot and OpenAI’s GPT-4 in two versions: emotional and rational. The emotional version used language and facial expressions to evoke positive emotions, while the rational version focused on presenting factual information in a straightforward manner. Subsequently, we conducted a laboratory experiment with over 100 participants, each interacting with either the emotional or rational version of Saleshat. The participants engaged in a sales conversation in which Saleshat made product recommendations about Bluetooth speakers, followed by inquiries about purchase likelihood and perceptions of the robot.

Our findings suggest that an emotional design of robot-based sales assistants may actually deter consumers from making purchases compared to a more rational design. Despite robots’ ability to display emotions and establish connections, consumers may prefer a more factual, straightforward approach when seeking product advice. Interestingly, this effect does not seem to be driven by different types of trust. These insights highlight the importance of considering not just technical capabilities but also social and emotional factors when designing robots for consumer interactions.

For managers and retailers considering the adoption of robot-based sales assistants, our research underscores the importance of balancing technical prowess with social and emotional considerations. While designing robots to appear more human-like may seem intuitive, it does not necessarily translate to increased purchase intent and consumer acceptance. From a broader societal perspective, our findings contribute to a more nuanced discussion about the benefits and risks associated with robot-based sales assistants. Despite prevalent concerns about new technologies, our study participants generally showed openness and even enthusiasm toward interacting with robots in a sales capacity, suggesting that the conversation surrounding robotic technology may need to be more balanced.