CausalMMM: Learning Causal Structure for Marketing Mix Modeling

TL;DR

A new causal MMM problem that automatically discovers the interpretable causal structures from data and yields better GMV predictions is defined that integrates Granger causality in a variational inference framework to measure the causal relationships between different channels and predict the GMV with the regularization of both temporal and saturation marketing response patterns.

Abstract

In online advertising, marketing mix modeling (MMM) is employed to predict the gross merchandise volume (GMV) of brand shops and help decision-makers to adjust the budget allocation of various advertising channels. Traditional MMM methods leveraging regression techniques can fail in handling the complexity of marketing. Although some efforts try to encode the causal structures for better prediction, they have the strict restriction that causal structures are prior-known and unchangeable. In this paper, we define a new causal MMM problem that automatically discovers the interpretable causal structures from data and yields better GMV predictions. To achieve causal MMM, two essential challenges should be addressed: (1) Causal Heterogeneity. The causal structures of different kinds of shops vary a lot. (2) Marketing Response Patterns. Various marketing response patterns i.e., carryover effect and shape effect, have been validated in practice. We argue that causal MMM needs dynamically discover specific causal structures for different shops and the predictions should comply with the prior known marketing response patterns. Thus, we propose CausalMMM that integrates Granger causality in a variational inference framework to measure the causal relationships between different channels and predict the GMV with the regularization of both temporal and saturation marketing response patterns. Extensive experiments show that CausalMMM can not only achieve superior performance of causal structure learning on synthetic datasets with improvements of 5.7%\sim 7.1%, but also enhance the GMV prediction results on a representative E-commerce platform.

Figures (10)

• Figure 1: The motivation of CausalMMM. (a) shows heterogeneous causal structures in MMM where nodes in different colors denote channel and target variables, respectively. (b) illustrates the saturation curve of marketing response determined by contextual factors, such as economy, events, etc..
• Figure 2: The overview of CausalMMM. The causal relational encoder predicts causal structures between marketing variables $\mathbf{X, y}$. The marketing response decoder learns to predict marketing variables given their past observations and contextual variables. This framework enables us to extract heterogeneous causal structures and learn marketing responses, simultaneously.
• Figure 3: The structure of the marketing response decoder. At each step, the decoder takes the inferred causal structure $\mathbf{z}$, the past observation of $\mathbf{X,y}$, and the historical hidden states $\tilde{\mathbf{h}}_{}^{t+1}$ as input to model marketing response.
• Figure 4: Causal structure learning performance (in AUROC) w.r.t. the number of shops $N$ ($d=10, T=120$).
• Figure 5: Visualization results for GMV prediction.