Native Advertisement Selection and Allocation in Social Media Post Feeds

We study native advertisement selection and placement in social media post feeds. In the prevalent pay-per-click model, each ad click leads to certain amount of revenue for the platform. The probability of click for an ad depends on attributes that are either inherent to the ad e.g., ad quality or related to user profile and activity or related to the post feed. While the first two types of attributes are also encountered in web-search advertising, the third one fundamentally differentiates native from web-search advertising, and it is the one we model and study in this paper. Evidence from online platforms suggests that the main attributes of the third type that affect ad clicks are the relevance of ads to preceding posts, and the distance between consecutively projected ads; e.g., the fewer the intervening posts between ads, the smaller the click probability is, due to user saturation. We model the events of ad clicks as Bernoulli random variables. We seek the ad selection and allocation policy that optimizes a metric which is a combination of i the platform expected revenue, and ii uncertainty in revenue, captured by the variance of provisionally consumed budget of selected ads. Uncertainty in revenue should be minimum, since this translates into reduced profit or wasted advertising opportunities for the platform. On the other hand, the expected revenue from ad clicking should be maximum. The constraint is that the expected revenue attained for each selected ad should not exceed its apriori set budget. We show that the optimization problem above reduces to an instance of a resource-constrained minimum-cost path problem on a weighted directed acyclic graph. Through numerical evaluation, we assess the impact of various parameters on the objective, and the way they shape the tradeoff between revenue and uncertainty.