算法推荐的营销效应与风险边界：信息茧房、过度刺激与消费者福祉

The Marketing Effect and Risk Boundary of Algorithmic Recommendation: Information Cocoons, Overstimulation, and Consumer Well-Being

摘要: 在数字经济背景下，电商平台营销正从流量导向转向长期价值经营，算法推荐在其中发挥关键作用。本文采用理论分析与机制推演方法，系统梳理了算法推荐在电商营销中的作用路径：个性化分发可降低消费者决策成本、改善匹配质量并促进信任累积；但也可能因目标偏置引发信息多样性下降与“信息茧房”，并在强刺激机制下诱发冲动消费，损害消费者自主性与长期福祉。基于此，本文提出以“可见、可控、可审计”为核心的平台治理框架，明确多样性底线、刺激强度上限、知情可控红线及弱势群体保护等风险边界，并从结果、过程与风险三个层面构建评估体系，以支持平台形成闭环改进机制，实现可持续营销。

Abstract: In the context of the digital economy, e-commerce platform marketing is shifting from a traffic-oriented approach to long-term value management, in which algorithmic recommendation plays a pivotal role. Using theoretical analysis and mechanism-based reasoning, this paper systematically clarifies the pathways through which recommendation algorithms influence e-commerce marketing: personalized distribution can reduce consumers’ decision-making costs, improve matching quality, and foster the accumulation of trust; however, it may also lead to reduced information diversity and “information cocoons” due to objective bias, and, under high-intensity stimulation mechanisms, trigger impulsive consumption that undermines consumer autonomy and long-term well-being. On this basis, the paper proposes a platform governance framework centered on “visibility, controllability, and auditability”, specifying risk boundaries such as a minimum standard for diversity, an upper limit on stimulation intensity, red lines for informed and controllable choice, and protections for vulnerable groups. It further constructs an evaluation system across outcomes, processes, and risks to support platforms in forming a closed-loop improvement mechanism and achieving sustainable marketing.