Scholarlink

In the age of digital marketing, firms must allocate advertising budgets efficiently by identifying which campaign strategies truly drive return on investment (ROI). This paper proposes a data-driven framework combining causal inference techniques with predictive modeling to assess and forecast the impact of campaign characteristics on marketing ROI. We employ different causal methods: Matching, Linear Regression, Inverse Probability Weighting (IPW), and Causal Forests to estimate the Average Treatment Effect on the Treated (ATT) of key features such as campaign type. These methods allow us to account for selection bias and reveal heterogeneous effects across subpopulations. Our findings confirm the negative impact of certain campaign types on ROI while highlighting the consistent positive effects of others. We further integrate these causal insights into a predictive model using SHAP (SHapley Additive exPlanations) values to quantify the contribution of each feature to the expected ROI of a given campaign. This research provides a robust, interpretable methodology to guide marketers in optimizing campaign design, tailoring strategies to specific audiences, and forecasting ROI with greater confidence.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Causal Inference and Predictive Modeling for Evaluating the Effectiveness of Online Marketing Campaigns

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Antoine Marmonier,
  • Carlos Monserrat Aranda

摘要

In the age of digital marketing, firms must allocate advertising budgets efficiently by identifying which campaign strategies truly drive return on investment (ROI). This paper proposes a data-driven framework combining causal inference techniques with predictive modeling to assess and forecast the impact of campaign characteristics on marketing ROI. We employ different causal methods: Matching, Linear Regression, Inverse Probability Weighting (IPW), and Causal Forests to estimate the Average Treatment Effect on the Treated (ATT) of key features such as campaign type. These methods allow us to account for selection bias and reveal heterogeneous effects across subpopulations. Our findings confirm the negative impact of certain campaign types on ROI while highlighting the consistent positive effects of others. We further integrate these causal insights into a predictive model using SHAP (SHapley Additive exPlanations) values to quantify the contribution of each feature to the expected ROI of a given campaign. This research provides a robust, interpretable methodology to guide marketers in optimizing campaign design, tailoring strategies to specific audiences, and forecasting ROI with greater confidence.