Title: Generative AI and Sales Productivity: Field Experiments in Online Retail

Abstract:

This study measures the immediate impact of Generative Artificial Intelligence (GenAI) on sales outcomes by conducting large-scale randomized field experiments. These trials involved millions of users and products on a prominent cross-border e-commerce platform. Between 2023 and 2024, the retailer embedded GenAI tools into seven distinct consumer-facing workflows, covering areas such as seller services, advertising, consumer-product matching, and customer support.

Our analysis reveals that GenAI integration boosts sales across most workflows. The magnitude of this effect varies from negligible to 16.3%, contingent upon the additional value GenAI provides over the company’s existing baseline practices. For the four specific applications that demonstrated positive sales results, the estimated annual incremental value is approximately $5. This figure represents a significant economic impact, particularly considering the retailer’s massive scale and the nascent stage of GenAI adoption.

The observed improvements are driven primarily by increased conversion rates rather than higher average cart values. This pattern aligns with the hypothesis that GenAI enhances the shopping experience by minimizing friction related to search, information retrieval, communication, and personalization. Crucially, these performance gains do not come at the cost of customer satisfaction; data shows no deterioration in post-purchase metrics, as both product return rates and customer ratings remain stable.

Furthermore, the study identifies significant heterogeneity in demand, noting that less experienced consumers benefit more substantially from GenAI implementation. These results offer new, large-scale causal evidence regarding the role of GenAI in enhancing sales productivity within the online retail sector, underscoring both its current tangible value and its wider future potential.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC