The composition of web services is a promising approach enabling flexible and loose integration of business applications. Numerous approaches related to web services composition have been developed usually following three main phases: the service discovery is based on the semantic description of advertised services, i.e. the functionality of the service, meanwhile the service selection is based on non- functional quality dimensions of service, and finally the service composition aims to support an underlying process. Most of those approaches explore techniques of static or dynamic design for an optimal service composition. One important aspect so far is mostly neglected, focusing on the output produced of composite web services. In this paper, in contrast to many prominent approaches we introduce a data quality perspective on web services. Based on a data quality management approach, we propose a framework for analyzing data produced by the composite service execution. Utilising process information together with data in service logs, our approach allows identifying problems in service composition and execution. Analyzing the service execution history our approach helps to improve common approaches of service selection and composition.