Algorithmic Targeting for Opaque Selling in Vertical Markets

Abstract

Motivated by algorithmic targeting and data management, we explore a scenario where the seller holds an advantage over consumers regarding match-related information about products. The seller optimizes a product line consisting of two vertically differentiated products alongside an opaque product resulting from their mixture, strategically recommending these products to potential consumers. We model algorithmic targeting using an information design framework, and our investigation revolves around understanding how algorithmic targeting shapes consumer purchasing behaviors and influences market equilibrium. Furthermore, we explore the potential orchestration between algorithmic targeting and opaque selling, facilitated by product-line design. These two closely related instruments coincide in ex-ante manipulating information while differing in their targeting objects. Interestingly, only when the basic products exhibit intermediate differentiation does the seller use both instruments. This is because, when the disparity between the two primary products is extreme (either too large or too small), algorithmic targeting makes opaque selling ineffective at increasing profits. However, when these differences are moderate, the two strategies can complement each other. Opaque selling enhances profitability by introducing intermediate product variety, enabling more nuanced market segmentation, while algorithmic targeting is more flexible in promoting the willingness-to-pay of a wider range of consumers. Furthermore, when conducting welfare analysis, the adoption of algorithmic targeting is found sometimes to reduce consumer surplus but can enhance overall social welfare, highlighting the need for careful regulatory oversight in this domain.