Causal Attribution and Evaluation Based on Large-Scale Advertising Data

Abstract

As the reach of digital advertising expands and multi-touchpoint data volumes surge exponentially, traditional advertising effectiveness evaluation methods based on correlation analysis can no longer accurately and effectively reflect the impact level from advertising touchpoints to behavioural conversion. This study constructs a causal attribution model and advertising effectiveness evaluation system based on causal principles, suitable for large-scale advertising data. Through latent outcome frameworks and causal effect estimation techniques, the model identifies the trajectory of advertising touchpoint influence. It further constructs an indicator system for causal effectiveness, primarily comprising average causal effects, behavioural uplift rates, and causal return on investment, establishing a systematic framework from causal modelling to effectiveness evaluation. The research achieves precise quantitative interpretation of advertising effectiveness, providing scientific grounds for formulating rational advertising allocation methods and enabling intelligent decision-making.