With the booming development of e-commerce, personalized recommendation systems have become the key to attracting users. However, traditional recommendation systems face user diversity and information overload, driving the search for more intelligent and adaptable solutions. Therefore, a multi-recommendation model for e-commerce is proposed by integrating backpropagation algorithm and genetic algorithm. The backpropagation algorithm is used to learn user historical behavior data and establish an initial recommendation model. Then, the genetic algorithm is applied to evolve and optimize the model to meet the personalized needs. Through continuous iteration and evolution, the recommendation model can better capture the potential interests and behavioral patterns of users. The proposed method reached the target value after 51 iterations, which was significantly faster than that of the pre-optimized model. The latter reached the standard after 100 iterations. In terms of error, the highest value was 43, with an average of about 15, while the highest value of the optimized network was only 40, with an average of about 12. Therefore, the optimized network performed significantly better than the pre-optimized network. The multi-recommendation mode based on optimized backpropagation algorithm performs better than traditional methods. This study provides a new approach for the design and optimization of e-commerce recommendation systems.