Digital Sales Assistants in ‘Bricks & Clicks’

Recent advancements in Large Language Models and Generative AI have opened new frontiers in retail shopper engagement through Digital Sales Assistants (DSAs). In this evolving tech landscape, retailers are challenged with optimizing DSAs for shopper benefit. While previous research on conversational agents has primarily focused on single features, DSAs integrate multiple features simultaneously to interact with shoppers. This necessitates a better relative understanding of the impact of DSA features when considered jointly. We surveyed 446 individuals using discrete choice experiments in a retail setting. Information origination and privacy aspects of a DSA drive shoppers' utility most, whereas non-verbal cues have less impact. However, the effects of non-verbal cues vary by context: In-store interactions thrive on humanized, credible cues, while at-home interactions succeed with less humanized, more likable DSA designs. Strong preference heterogeneity unveiled only a weak link between DSA features and better recommendation adherence.