The quality of cross-cultural marketing is related to the prospect of business development. This paper analyzes the challenges encountered by international companies in cross-cultural marketing. A two-stage feature selection algorithm based on information gain and Pearson’s correlation coefficient (IG-PPMCC-PCA) is proposed to optimize sales feature dimensions in combination with principal component analysis. A causal inference-driven gain model is constructed to predict the incremental user transaction willingness under marketing intervention. The results show that the gain model in this paper performs well in dataset training, with a mean square error of 78.6460, an absolute error of 1.51%, and AUUC and Qini reaching 0.8365 and 0.1295, which are better than the other five comparison models. Under the cost constraints, the model in this paper reaches the user group bringing 8.34% more than randomly selecting users. The intelligent marketing gain model that integrates feature selection and causal inference can significantly improve the accuracy and cost-effectiveness of cross-cultural marketing decisions.
