The rational allocation of marketing resources is the focus of enterprises to increase economic returns. In this paper, we use maximum-minimum, zero-mean and fractional calibration normalization to normalize the enterprise customer data preprocessing and improve the clustering generalization performance. Customers with different characteristics are clustered by K-means algorithm to mine the marketing concerns of similar customers. Combine with the multi-item benefit evaluation formula to calculate the user fitness score and obtain the maximized marketing benefit. Construct the marketing resource model based on information entropy to quantify the marketing resource allocation system loss in order to select the optimal decision variables. The omni-channel marketing resource allocation optimization process of cultural and creative e-commerce stores is exemplified to analyze the effect of this paper’s method. The results show that the best classification effect is obtained by clustering customers into six classes and analyzing their corresponding characteristics. After performing the marketing resource optimization, the final product price decreases to RMB 36 per piece, the total sales volume increases to 150*104 pieces, the market share increases to 40%, and the profit improves to 16.1%.
