Dynamic budget allocation for social media advertising campaigns: optimization and learning

This paper suggests a method for optimizing a dynamic budget allocation policy for an advertising campaign posted through a social network (e.g., Facebook, Instagram). The method, which considers unique features of social network marketing, yields an optimal targeted budget allocation policy over time for a single ad campaign and minimizes the campaign's length, given a specific budget and a desired level of exposure of each marketing segment. The model incorporates a general ‘effectiveness function’ that determines the relationship between the value of an advertising bid at a given time and the number of newly exposed users at that time. We develop closed-form solutions for dynamic budget allocation for several forms of the effectiveness function. We apply the approach to data obtained from a real-life ad campaign and show how a curve fitting regression procedure can estimate the shape and the parameters of the effectiveness function. Numerical simulations show the extent to which the optimal advertising policy is sensitive to the problem parameters.