With the rapid iteration of digital marketing and the total online advertising spending of Chinese enterprises exceeding one trillion yuan, approximately 42% of enterprises face difficulties in accurately measuring advertising effectiveness and budget allocation due to the inability of traditional attribution models to address the misjudgment of touchpoint contribution in the fragmented user journey. Furthermore, existing digital advertising attribution methods often rely on association analysis, confusing correlation with causation, leading to distorted quantification of touchpoint contribution, unreasonable budget allocation, and low ROI, failing to meet the core requirements of digital marketing. To this end, this paper first reviews the current research status and shortcomings of advertising attribution and budget allocation in the context of digital marketing. Then, it constructs a multi-touchpoint attribution model based on causal inference, integrating methods such as difference-in-differences and causal forests to accurately quantify the true causal effects of each advertising touchpoint. Next, it designs a dynamic budget allocation optimization algorithm based on attribution results to achieve intelligent budget scheduling. Finally, it builds a complete system architecture to complete model training, effect verification, and deployment. Experiments show that the attribution accuracy of Causal-AA reaches 89%, which is 18, 22, and 15 percentage points higher than the last-click model, first-click model, and linear model, respectively, and reduces customer acquisition costs by 20%-30%, verifying the effectiveness and practicality of the model and system.
Ruiqiang Li (Thu,) studied this question.