Nodding of virtual humans does not necessarily increase customer trust: a temporal analysis

Virtual humans can potentially facilitate engaging customer interactions, but remain rarely used in customer service. Their embodiment enables virtual humans to use non-verbal listening cues like nodding. Nodding is theorized to increase trust in virtual humans, but literature on this effect is scarce, inconsistent, and only uses measurements after interactions. This study therefore aims to research how nodding of virtual humans influences customer trust, by (1) studying the following hypothesized mediation: nodding (social cue) triggers perceived responsiveness (social signal), which subsequently increases trust (social response); and (2) studying temporal trust dynamics within a customer interaction using exploratory analyses. An online experiment ( N = 429) was performed, using a between-subjects design with two conditions: a virtual human either nodded during listening or did not. To isolate nodding optimally, participants watched a video of a customer interaction: a virtual human advised a customer on a wardrobe. While watching, participants used a slider scale to keep updating their temporal trust in the virtual human. Afterwards, participants answered questionnaires on trust and perceived responsiveness. Results showed that nodding did not increase trust or perceived responsiveness, nor did nodding build temporal trust. Temporal trust was built throughout the customer interaction, mostly when the virtual human was speaking. Exploratory analysis found that nodding increased anthropomorphism. Possibly, nodding did not influence trust because other social cues already caused trust-building. Virtual human developers should therefore be aware that nodding does not necessarily build customer trust. The temporal trust scale was useful for visualizing trust-building throughout customer interaction videos.