关键词: 信息设计, 在线购物平台, 策略推荐算法, 贝叶斯劝说

Abstract: Recommendation is being widely adopted by online shopping platforms to enhance consumer purchasing behavior. Given that different recommendation strategies exert varying effects on consumer behavior under different prior beliefs, this study explores how the firm and platform should design their information strategies to maximize profits. We construct a multi-stage non-cooperative game model using Stackelberg dynamic game theory for backward induction, incorporating reputation costs under the platform’s revenue-sharing contract. The analysis examines whether the firm should adopt the recommendation feature, the optimal intensity of strategic recommendations to influence consumer beliefs, and the platform’s optimal access fee for the feature. The results show that the recommendation feature expands the consumer base and stimulates purchases by reshaping consumer beliefs, yet even when the platform waives access fees, high-brand-equity firms may lack adoption incentives. A moderate level of strategic recommendations can boost sales, with this effect contingent on the feature’s accuracy and consumer beliefs. Under asymmetric information, the firm can mitigate profit extraction by the platform, and prior negotiation may avert a "lose-lose" outcome. The study provides quantitative analysis and managerial insights to support these findings.

Key words: information design, online shopping platform, strategic recommendation algorithm, Bayesian persuasion