Causal-driven attribution (CDA): Estimating channel influence without user-level data

2025-12-24

AI Summary

Facing stringent privacy regulations that impede access to user-level click-path data, this paper proposes a novel marketing attribution paradigm relying solely on aggregated exposure data. Methodologically, we introduce the first framework integrating temporal causal discovery (PCMCI) with structural causal models (SCM), enabling cross-channel dependency modeling and counterfactual effect estimation without user identifiers or path tracking. The framework ensures both interpretability and robustness: on synthetic data, it achieves relative RMSEs of 9.50% and 24.23% for recovering the true causal graph and predicting the causal graph, respectively—substantially outperforming existing aggregate-level attribution methods. To our knowledge, this work provides the first theoretically sound and empirically verifiable causal attribution solution for privacy-sensitive marketing analytics.

Technology Category

Machine Learning: Causal LearningReasoning under Uncertainty: CausalityData Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust

Application Category

User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertisingSecurity and Privacy: Large-scale security measurementsSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Attribution modelling lies at the heart of marketing effectiveness, yet most existing approaches depend on user-level path data, which are increasingly inaccessible due to privacy regulations and platform restrictions. This paper introduces a Causal-Driven Attribution (CDA) framework that infers channel influence using only aggregated impression-level data, avoiding any reliance on user identifiers or click-path tracking. CDA integrates temporal causal discovery (using PCMCI) with causal effect estimation via a Structural Causal Model to recover directional channel relationships and quantify their contributions to conversions. Using large-scale synthetic data designed to replicate real marketing dynamics, we show that CDA achieves an average relative RMSE of 9.50% when given the true causal graph, and 24.23% when using the predicted graph, demonstrating strong accuracy under correct structure and meaningful signal recovery even under structural uncertainty. CDA captures cross-channel interdependencies while providing interpretable, privacy-preserving attribution insights, offering a scalable and future-proof alternative to traditional path-based models.

Problem

• Estimates channel influence without user-level data
• Uses aggregated impression data for privacy compliance
• Quantifies causal channel contributions to conversions

Innovation

• Causal-driven attribution uses aggregated impression-level data only
• Integrates temporal causal discovery with structural causal modeling
• Provides privacy-preserving attribution without user-level tracking