DescriptionBusiness to Business (B2B) marketing aims at meeting the needs of other businesses instead of individual consumers, and thus entail management of more complex business needs than consumer marketing. The buying processes of the business customers involve series of different marketing campaigns providing multifaceted information about the products or services. While most existing studies focus on individual consumers, little has been done to guide business customers due to the dynamic and complex nature of these business buying processes. In this dissertation, we focus on providing data-driven solutions to achieve two important business goals: reduce the buying cycle time and increase the conversion rate. Specifically, we first introduce a unified view of social and temporal modeling for B2B marketing campaign recommendation to reduce the buying cycle time. Along this line, we exploit the temporal behavior patterns in the buying processes of the business customers and develop a B2B marketing campaign recommender system. Specifically, we first propose the temporal graph as the temporal knowledge representation of the buying process of each business customer. We then develop the low-rank graph reconstruction framework to identify the common graph patterns and predict the missing edges in the temporal graphs. In addition, we also exploit the community relationships of the business customers to improve the performances of the graph edge predictions and the marketing campaign recommendations. Results from extensive empirical studies on real-world B2B marketing data sets show that the proposed method can effectively improve the quality of the campaign recommendations for challenging B2B marketing tasks. Furthermore, we develop two different approaches aiming at improving the conversion rate. We first present a novel unified framework to integrate two important marketing tasks--customer segmentation and buyer targeting. Instead of combining these two tasks in a simple step-by-step approach, we formulate customer segmentation and buyer targeting as a unified optimization problem. Then, the customer segments are adaptively realized during the targeting optimization process. In this way, the integrated approach not only improves the buyer targeting performances but also provides a new perspective of segmentation based on the buying decision preferences of the customers. Finally, we introduce a predictive lead scoring model which can help sales representatives to identify prospective leads from a large pool of candidates in a B2B environment. Specifically, we provide a multi-focal lead scoring framework which can improve the performance of predictive lead scores. However, independent modeling at focal level would be problematic for segments with few representative samples. We use the Multi-Task Learning framework to address this problem by exploiting commonalities shared by focal groups and automatically balancing between unification of all groups and individualization of each group. Therefore, such a multi-focal tailored lead scoring model gives a better insight into factors influencing the conversion of leads.