Robot Sales Assistants

Advances in technology, tech-savvy consumers, and the new challenges posed by the COVID-19 pandemic are urging retailers to innovate and enhance the customer experience. In this setting, the application of service robots is rapidly spreading across several store types, from grocery stores to luxury ones, performing more complex tasks and interacting with customers. Retailers are striving to identify the right balance between robot and human sales assistants, while at the same time wondering how robots can affect in-store consumer behavior. Through a neuro-scientific approach, this study evaluates the degree of immersion of human-human vs. robot-human in-store interactions across the stages of the selling process and how this affects store visit length.

The market of service robots is growing rapidly and robots are becoming more affordable and technologically advanced. In retail, robots were traditionally introduced to handle basic tasks and inventory. However, in recent times retailers have started employing service robots to innovate customer interactions. Past research has focused on evaluating human-robot interactions (HRI) in the retail environment using qualitative methods, and no attention has been given from a neuroscientific perspective.

In service intensive stores, the sales assistant-customer interaction follows the selling process that is a script of several consequent steps, which generates customer immersion and predicts customer attention and engagement. Immersion is a neurological and psychological process, related to narrative transportation, which occurs when a person experiences a deep attentional and emotional involvement with the environment. It can be measured through the software platform ImmersionNeuroTM, which captures in real-time neural signals associated with attention (increased heart rate and electrodermal activity) and vagal tone (increased heart rate variability).

In this study, we assessed how sales assistance mode, i.e. human-robot interactions (HRI) versus human-human interactions, affects the relationship between immersion and dwell time. Dwell time is defined as the amount of time spent inside the store by customers. Past research suggests that longer visits to a store increase the opportunity to interact with the store environment and merchandise, which in turn could lead to higher purchase intention, conversion rate and profitability (Newman et al. 2006; Mallapragada et al., 2016; Sorensen et al. 2016). Specifically, we focused on four different stages of the selling process (welcome, store introduction, brand storytelling, and surprise) and we measured the interactions through wearable sensors connected to the software platform. We proposed that customers generally respond to the presence of a service robot by increasing their level of engagement in the shopping experience. However, we also hypothesized that the degree of immersion in the shopping experience would be significantly higher when interacting with a human sales assistant. Moreover, we posited that the impact of immersion on dwell time is weaker when interacting with a robot sales assistant compared to with a human sales assistant as customers have been found by previous studies to prefer human labor when examining products with higher symbolic value like fashion items (Granulo et al., 2020).

Methods

Participants (N=50, 28F, 22M, ages 18-65, average age 37.4) participated in a field experiment conducted in a retail store selling Italian leather goods. Each participant was given $20 and randomly allocated to an experimental condition, i.e. interaction with a robot sales assistant versus a human sales assistant. Biometrics were collected through the wearable sensors and sent to the software platform.

During the experiment, we used the social robot, Jibo. Given that robots are only flexible within defined limits, we coded the robot to interact with the customers in the four aforementioned stages of the selling process. As service robots have a codified knowledge and a systematic setting, we trained the sales assistant to memorize the service script and to reproduce the same words and behaviors of the robot. All interactions were recorded by an external HD camera positioned in the store and connected through a Bluetooth microphone and intercom placed on participants to record the voice and reduce the environmental noise during the recording; Participants were not given a specific length of time to complete the experience, but were left free to browse in-store. Each experience was timed to obtain dwell time data which are reported here in seconds. After completing the shopping experience, participants were presented with a survey collecting demographics and familiarity with Italian luxury brands. After the study, ECG artifacts were manually removed from data and R-R intervals were identified and extracted through ImmersionNeuroTM. Upon completion of the study, selling process stages were coded and video recordings were paired and synchronized with the biometric data. Data were then analyzed through ANOVA and regression analysis using Hayes’ PROCESS model 1 with 5,000 bootstrapping5.

Key findings

1. With reference to the entire shopping experience, we assessed a significant difference in immersion across the two experimental conditions (Mhuman= 4.62, Mrobot= 4.75, F=16.41, p<.001). We found that immersion was significantly higher in the robot condition during the surprise moment (Mhuman= 4.63, Mrobot= 4.90, F=12.59, p<.001) while there were no significant differences across the other moments of the selling process.

2. To evaluate the impact of immersion on dwell time, we conducted a multiple regression analysis including age, gender, and familiarity with Italian brands as control variables. Immersion has a significant and positive effect on dwell time (ß=.178, p<.001). This confirms prior research as the more consumers are engaged in a shopping experience, the more likely they are to spend more time in-store.

Figure 1: Effect of immersion on dwell time across sales assistance mode and stages of the selling process

3. Through a moderation analysis, we assessed the effect of the interaction with a human vs. robot sales assistant on the relationship between immersion and dwell time. The interaction between immersion and sales assistance mode (human vs. robot) is statistically significant in the welcome (F= 18.65, p<.001), store and merchandise introduction (F= 19.56, p<.001), brand storytelling (F= 10.82, p=.001), and surprise (F= 8.14, p=.004) moments. Participants exposed to a robot sales assistant generally reported a shorter dwell time compared to those who interacted with a human sales assistant (Figure 1). Thus, the positive effect of immersion on store visit length is weakened by the introduction of a service robot sales assistant.

Discussion

The results demonstrate that there are important differences between human-robot interactions (HRI) and human-human interactions in a retail store. We generally observed that the presence of a robot increased the level of immersion in a shopping experience. Customers also showed a deeper connection and higher level of engagement with the shopping experience when interacting with a robot. The level of attention placed in the interaction increases, which in turn enhances the overall enjoyment stemming from the shopping experience. This effect is particularly significant in the surprise phase, in which a small and unexpected gift was offered to customers, who showed a higher degree of engagement when a robot was performing this specific task compared to a human sales assistant.

Another relevant finding of this study is related to the relationship between immersion and dwell time. Dwell time is an important metric to monitor in retail spaces as an increase in customer dwell time could lead to a higher purchase likelihood and positively affect impulse buying behavior. Our study confirms that immersion positively affects dwell time. In other words, the more customers are deeply engaged in the shopping experience, the more they are likely to spend additional time in the store. However, we observed that this relationship is significantly stronger when customers are interacting with a human sales assistant compared to a service robot. Those who showed high levels of immersion and were interacting with a human sales assistant during the brand storytelling moment stayed in-store longer (18 min 16 sec), followed by those who were introduced to the store by a human sales assistant (17 min 39 sec). Dwell time when interacting with the robot sales assistant was significantly lower, with the longest visits recorded for those who were deeply immersed in the experience and interacted with the robot in the brand storytelling (13 min 17 sec) stage and those who showed low immersion and were introduced to the store by the robot (12 min 41 sec).

Potential applications in retail

1. The presence of a robot has positive effects on customers during the service encounter. Although the communications are artificial, customers enjoy having a conversation with service robots and reward it with higher levels of attention and engagement. Therefore, we recommend retailers to consider the introduction of conversation robot types in the store to enhance the service experience.

2. However, the role of human sales assistants is still key to increasing dwell time and keeping customers in-store, increasing their likelihood to purchase or impulse buying. This result can be explained by the importance of social components in human relationships, rather than the simple transactional element. Hence, the presence of a human sales assistant is still necessary for developing a healthy relationship with customers.

3. Immersion increases dwell time during welcome and brand storytelling, highlighting the convergence of retail and neuroscience. As retail stresses the importance of welcoming and brand storytelling during the service encounter, biosensors show that activity in the peripheral physiology occurs in response to emotional and persuasive stages of the selling process. Both stages can be facilitated by humans or robots, but only humans generate higher permanence in the store.

4. High immersion levels during the surprise moment demonstrate that when sales assistants engage in “extraordinary behaviors”, shoppers boost their emotional arousal and shopping enjoyment. Receiving a surprise from a robot results in a deeper immersion in the experience. This occurs because it is less expected as it is widely known that tech follows a script. So, retailers could consider a wider application of robots in their surprise-and-delight strategies.

About the authors:
Gaia Rancati is a Professor of Marketing and Neuromarketing at Allegheny College and IULM University (Italy). Isabella Maggioni is a Professor of Marketing at ESCP Business School.

This was originally published in Insights Magazine, NMSBA members have access to the full archive of this quarterly magazine on neuromarketing. Interested in joining? Check the options.