Solution design has been more of an art than an engineering discipline. Lots of researchers and practitioners have proposed and exercised different kinds of approaches with varied success. Most of these methods seem to have focused on building new solutions from scratch. However, enterprise solutions today are mostly built on top of an existing IT infrastructure. The notion of SoA is trying to pave a way to integrate heterogeneous components together to meet new business needs. When a new requirement is given to a system developer in the form of business processes, it would be ideal if s/he can make the best of existing services for many reasons. In this paper we propose a data driven approach to provide service composition guidance to implement the given requirement. Based on the relations among business domain data and service domain data, we can generate additional data mediations according to three composition rules. With these data relations and composition rules, we give a formal approach to devise choreography of services from current service portfolio, plus additional data mediation artifacts to realize a given requirement. Our work can be seen as an effort to bridge the gap between business and service domain.