In our evolving computing environment with heterogenously distributed information systems, products are continuously modified and changed. During this process a change to one part will, in most cases, results in changes to other parts. Therefore, in design and redesign for customization, predicting this change presents a significant challenge. Changes are required to fix faults or to improve or update products. This paper reports on the development of a change impact analysis factor adaptation model, a fault and failure assumption model and the implementation of a generic change propagation framework for evaluating and assessing utility service provisioning in a Grid service environment. While implementing the framework, data was collected for a period of 3 years which helped in predicting an immediate year. The obtained results from our pre- diction shows the framework, its associated models and Bayesian statistics as satisfying the criteria for a significant prediction accuracy in evaluating and assessing the effect of a change of service in a grid environment when compared to an unreported regression method.