Effect of Complex Multimedia Advertising Campaigns : A New Automated Method for Big Data

The proliferation of online advertising not only brings new opportunities for advertisers to reach the consumers but also allows them to efficiently track large-scale individual-level data on ad exposures and conversions. Moreover, ad campaigns leveraging both online and offline media platforms are becoming an increasingly popular option. However, advertisers often question whether multimedia campaigns are actually effective and it remains unclear how to maximize their success. This study introduces a novel tree-structured causal inference model, which is nonparametric, flexible, computationally efficient, and suitable to analyze complicated nonlinear effects. Both the simulation and empirical studies demonstrate the effectiveness of the model to correct for selection bias and estimate the unbiased causal effects of complex multi-dimensional ad treatments. Furthermore, the model enables automatic segmentation of consumers, allowing advertisers to better optimize the allocation of ad resources via accurate targeting. To the best of our knowledge, it is the first automated framework that performs end-to-end causal estimation and segmentation analysis for multimedia marketing based on big data