A growing number of retailers offer products/services via the Internet as well as at brick-and-mortar stores. These multiple marketing channels increase the opportunity for customers to buy. In addition, retailers can analyze detailed customer information gained from multi-channel shopping in order to improve their service quality. In view of these facts, we develop a specialized recommender method based on the weighted non-negative matrix factorization to expand the number of customers that make use of multiple marketing channels. Our method effectively uses the similarities of customers and products to promote customers' multi-channel shopping. Computational results demonstrate that our method delivers better recommendation performance than the existing recommender methods.