Abstract: Augmented reality (AR) has emerged as a significant technological innovation in e-commerce. While many studies have investigated its effects on consumer perceptions and purchase intentions, its overall impact on post-sales product return remains unclear. Drawing on dual process theory and cue utilization theory, this study introduces a novel concept of contextual experience to investigate AR’s due impact on product returns. Specifically, AR can reduce pre-purchase fit uncertainty, particularly regarding the contextual experience of the products, thereby enabling more informed purchases and thus lower return rates. However, AR’s vivid and salient visual cues may also bias consumer judgements: consumers may overemphasize these cues in their decisions and neglect other non-contextual product attributes. Such bias can increase purchase likelihood but also result in higher return rates if actual experiences fail to meet expectations, potentially causing losses for retailers. We test these mechanisms through a multi-method approach: a quasi-experimental analysis of e-commerce data and a controlled online experiment. The results show that AR reduces returns for products with high contextual experience relevance but increases returns for products with low relevance. In addition, products with more accessible non-contextual attribute information are less likely to experience return rate increases following AR adoption. Our findings highlight AR’s dual role in consumer decision-making and suggest that retailers should tailor AR implementation to product characteristics and information availability rather than adopting a uniform strategy. This study is also among the first to evaluate AR's impact on product returns.

Abstract: Online platforms face two persistent challenges in reducing consumer search friction: inaccurate query formulation and the inherent trade-off between precise search and product discovery. Most refinement tools simply narrow the search results based on consumer-specified queries, which, if imprecise, could yield unsatisfactory results. At the same time, overly precise, frictionless searches, while helping consumers identify the target product quickly, may discourage exploration, limiting potential sales opportunities. One promising solution lies in AI-driven tools like product-knowledge-graphs-powered search (KGS). KGS structures product information into subcategories, thereby guiding effective and efficient query formulation for consumers. By simultaneously priming consumers with diverse options and supporting precise search refinement, KGS can enhance both exploration and efficiency. Despite its potential, limited research addresses the value of KGS for consumers and the platform, particularly regarding consumers with varying search intents. Through a randomized field experiment conducted on a leading on-demand service platform, this study quantifies the effects of KGS on consumer search behavior and outcomes. Our results demonstrate that KGS significantly enhances consumer exploration and engagement: KGS leads to more frequent and diverse query submissions, boosting both clicks and orders within a search session. These effects are most pronounced when consumers face high uncertainty in their searches, especially when consumers initiate searches with broad, ambiguous queries. Moreover, the increases in clicks and purchases are driven primarily by hedonic goods and by novice platform users, who often lack knowledge about available products on the platform and platform navigation. This research provides the first empirical evidence of the value of KGS in the consumer search context, offering key insights for platform managers on balancing efficient search with guided product discovery - two often competing objectives in online search.

Abstract: Online retailers increasingly integrate augmented reality (AR) into product pages to provide additional information on how the item fits consumers’ physical context. However, the click-through rates for AR suggest that sometimes consumers bypass these tools, relying instead on other available information. Through modeling consumers’ information acquisition and purchase decisions, we derived propositions that are further supported by real-world observations. Our findings reveal the key relationships that a higher click rate to AR tools is associated with (i) lower volume of other information; (ii) higher valence of product ratings; and (iii) higher relative importance of contextual fit. With a quasi-experiment setting, we further find that online rating characteristics moderate the effect of AR on product sales, and the average rating score increases after the introduction of AR. In conclusion, this study highlights the boundary conditions of AR adoption effects, delivering both theoretical contributions and managerial implications to online business operators.