Most of existing approaches consider composition optimization as a back-end process and rarely expect non-technical end-users to interact directly with the optimization process. Thus these approaches do not enable the end-user to leverage the effects of direct interaction of the process of composition. Indeed, corrective actions can be taken only before and after the execution, not during composition construction. We focus on end-user support and optimized service selection during composition modeling. Of particular interest hereby is whether business users with limited IT skills can leverage our approach to optimize and consume web services and their composition. © 2012 IEEE.