With the rapid development of cross-border e-commerce, product selection has emerged as one of the crucial factors determining the success or failure of enterprises in the cross-border e-commerce domain. This study employs a combination of qualitative and quantitative research methods, taking the XGBoost algorithm as the core and integrating data mining, machine learning, and other technical means to construct a new cross-border e-commerce product selection model. The specific technical route encompasses steps such as data collection and pre-processing, feature selection and extraction, model construction and optimization, and model evaluation and comparative testing. The results indicate that the cross-border e-commerce product selection model based on big data and XGBoost, in comparison with traditional methods, has high accuracy and practicality and is a more scientific, efficient, and intelligent approach to cross-border e-commerce product selection. It can effectively assist enterprises in enhancing their product selection efficiency and success rate.