Location-based social network has received much attention recently. It provides rich information of social and spatial context for researchers to study users’ behaviors from different aspects. A number of recent efforts focus on recommending locations, users, activities, and social medias for users. Unlike previous works, we intend to make recommendations for vendors, assisting vendors in finding potential customers in location-based social network. We propose a framework to recommend potential customers to vendors (called RPCV) in location-based social network effectively and efficiently. To find the best set of customers, RPCV takes both spatial relations and user preference into consideration. A reverse spatial-preference kRanks algorithm, which effectively combines spatial relations with user preference, is also proposed. Our experimental results on real datasets from Foursquare and Brightkite show that our framework has higher performance than other state-ofthe- art approaches.