Abstract: In recent years,China’s platform economy has developed rapidly,and its position and role in the overall economic and social development have become increasingly prominent.As a core element of the platform economy,the intrinsic relationship between Internet traffic and the platform economy needs to be examined.Internet Celebrity Economy/Influencer Economy refers to the phenomenon where creators/influencers on video-sharing and live-streaming platforms leverage their fame or talents to attract followers and then monetize the traffic through collaborations with businesses.The high loyalty,strong stickiness,and long retention time of the streamers’ fans significantly differentiate them from casual viewers.The direct competition among streamers underscores the importance of analyzing the oligopolistic nature of traffic entry points. An interesting fact is that manufacturers of different categories often adopt different live-streaming e-commerce marketing models.Some product promotion links seem to be everywhere.Some merchants place advertisements on all platforms and multiple channels without distinguishing channel attributes.There even appears a contradictory situation where game zone anchors and knowledge zone anchors recommend the same product at the same time.Some merchants also place advertisements on multiple channels,but mainly in the form of pre-sales or flash sales,with a small amount of supply and no large-scale promotion.Other merchants focus on highlighting the “channel exclusive” feature in promotion,with limited release as the main selling point. This paper argues that the differentiation of marketing models is related to product categories to a certain extent.Unlike registered users of traditional platforms,potential consumers in the field of live-streaming e-commerce often exist in the form of fans of anchors or brands.Fan attribute is the most distinct personal characteristic of the audience of live-streaming sales.The “die-hard” audience of anchors or brands stay in the live-streaming room frequently and for a long time,with high product exposure frequency,high trust in anchors or brands,and high possibility of accepting recommendations; The “passer-by” audience of anchors or brands do not have a strong awareness of the anchors or brands.They mostly enter the live-streaming entrance through the way of enterprise paid promotion of the live-streaming room,with a general stay time,and the possibility of accepting recommendations depends on the cost-effectiveness of the product; The “haters” of anchors or brands have a strong dislike for the anchors or brands,and it is almost impossible for them to become live-streaming audiences,let alone accept recommended marketing.The acquisition costs of these three types of users show significant horizontal differences.Hence,the recognition of commodity value by these three types of users,that is,the correlation between commodity preference and live-streaming audience,is the key to distinguish commodity categories. Hence,this paper develops a duopoly model of platforms based on the Hotelling model,which allows firms to price traffic.It finds that when consumers’ preferences for goods are completely independent of their channel loyalty,firms always choose bilateral traffic diversion,and both partial and full market coverage are possible.When consumers’ preferences for goods are completely negatively correlated with their channel loyalty,the only possible scenario is bilateral traffic diversion by firms with partial market coverage.When consumers’ preferences for goods are completely positively correlated with their channel loyalty,two scenarios may occur:bilateral traffic diversion with full market coverage,or unilateral traffic diversion with partial market coverage. Subsequently,this paper analyzes the impact of changes in platforms’ customer acquisition costs and market segmentation on social welfare.It provides targeted policy recommendations for various markets based on differences in consumer types:when consumers’ preferences for goods are completely independent of their channel loyalty,the degree of market coverage can be used as an intuitive indicator of welfare.Markets with higher coverage generally have better consumer welfare properties without losing too much total social surplus.When consumers’ preferences for goods are completely negatively correlated with their channel loyalty,central planners can restrict behaviors such as induced sharing and forwarding,and penalize false or exaggerated elements in promotional activities to reduce the willingness of undesired users to participate in marketing activities,thereby preventing firms from exiting the market.When consumers’ preferences for goods are completely positively correlated with their channel loyalty,central planners should control channels’ customer acquisition costs as much as possible to promote more transactions and should pay more attention to markets with lower segmentation,especially being vigilant against practices such as “big data price discrimination” that harm consumer rights. This research not only has positive implications for guiding firms on how to choose appropriate traffic channel layouts but also fills a gap in the existing literature regarding the impact of the correlation between consumer preferences and traffic channel loyalty on market equilibrium.It provides a new perspective and explanatory framework for the theoretical study of platform economics.