Scholarlink

This study investigates the effectiveness of digital marketing channels in driving sales conversions by combining causal inference methods Randomized Controlled Trials (RCT), Propensity Score Matching (PSM), and Difference-in-Differences (DiD) with predictive modeling. The results from RCT and PSM indicate no statistically significant difference between the control and treatment groups. However, the DiD analysis reveals that the treatment group outperformed the control group by 2.47%, highlighting the effectiveness of certain channels. Predictive modeling was conducted using Random Forest and Logistic Regression. The Random Forest model achieved superior performance with an accuracy of 89.69%, precision of 90.21%, and recall of 99.00%. In contrast, Logistic Regression obtained a lower accuracy of 87.94% but exhibited high recall, demonstrating its strong ability to detect converting customers. The Return on Investment (ROI) analysis underscores significant performance variation across marketing channels, necessitating a more strategic budget allocation. The findings offer actionable insights for optimizing campaign strategies and improving ROI. This research also addresses the growing complexities in evaluating digital marketing effectiveness due to emerging trends such as AI-driven analytics, cross-device tracking, and privacy-centric regulations. By integrating causal inference with predictive modeling, the study bridges gaps in attribution analysis within multi-channel environments and provides scalable strategies for adaptive marketing decision-making.

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

Evaluating the Effectiveness of Digital Marketing Channels: A Causal Inference and Predictive Modeling Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Deden Witarsyaha,
  • Yoga Vikriansyah Wijaya,
  • Ririn Oktavianiva

摘要

This study investigates the effectiveness of digital marketing channels in driving sales conversions by combining causal inference methods Randomized Controlled Trials (RCT), Propensity Score Matching (PSM), and Difference-in-Differences (DiD) with predictive modeling. The results from RCT and PSM indicate no statistically significant difference between the control and treatment groups. However, the DiD analysis reveals that the treatment group outperformed the control group by 2.47%, highlighting the effectiveness of certain channels. Predictive modeling was conducted using Random Forest and Logistic Regression. The Random Forest model achieved superior performance with an accuracy of 89.69%, precision of 90.21%, and recall of 99.00%. In contrast, Logistic Regression obtained a lower accuracy of 87.94% but exhibited high recall, demonstrating its strong ability to detect converting customers. The Return on Investment (ROI) analysis underscores significant performance variation across marketing channels, necessitating a more strategic budget allocation. The findings offer actionable insights for optimizing campaign strategies and improving ROI. This research also addresses the growing complexities in evaluating digital marketing effectiveness due to emerging trends such as AI-driven analytics, cross-device tracking, and privacy-centric regulations. By integrating causal inference with predictive modeling, the study bridges gaps in attribution analysis within multi-channel environments and provides scalable strategies for adaptive marketing decision-making.