Ahn, Hyun-Soo; Ryan, Christopher Thomas; Uichanco, Joline; Zhang, Mengzhenyu; (2025) Valuing Influence with Social Learning. Manufacturing & Service Operations Management 10.1287/msom.2025.0348. (In press).

Preview

Text Zhenyu_Paper_3___Valuing_Influence_MSOM_manuscript__Accepted_version_.pdf - Accepted Version Download (1MB) | Preview

Abstract

Problem definition: Influencer marketing has become a prevalent strategy to promote products through social media. This paper examines the value of influencer marketing when followers not only learn from the influencer’s signal but can also engage in social learning by observing peers’ purchase behaviors and reviews. Methodology/results: We adopt an information design framework to analyze how a firm should value an influencer based on two key dimensions: the accuracy of the influencer’s past recommendations (informativeness) and the extent to which followers rely exclusively on the influencer versus learning from peers (charisma). Managerial implications: Our model uncovers insights about the interaction between information design and social learning. First, the naive intuition that the influencer is less valuable with social learning does not always hold. The influencer holds greater value under the social learning context when customers have a moderate intention to buy as her endorsement reinforces customer convictions, making them resilient against later negative feedback from other followers. Second, when the firm can strategically select an influencer, the optimal information structure is biased toward the positive signals: always endorse good products (true-positive rate of one) but sometimes endorse bad products (nonzero false-positive rate). Third, the optimal influencer when social learning exists has a lower false-positive rate than the one without social learning, meaning that when there exists subsequent social learning, it becomes even more important to have an influencer whose positive endorsement is trustworthy. In other words, the optimal influencer should be able to reveal more information with social learning than without social learning. // Funding: This work was supported by the National Science Foundation [Grant 2208189]. // Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2025.0348.

Download activity - last month

Export as XMLJSONCSV

Download activity - last 12 months

Export as XMLJSONCSV

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item