Developing dependable complex applications using IoT and cloud services is very challenging. Using service APIs and client libraries the developer can glue various software capabilities to build complex IoT Cloud applications but the developer also needs to arbitrarily extend and model functional and quality aspects of new components, connectors, and their interactions. Hence, knowledge about existing IoT Cloud and modeling is crucial. However, due to the lack of knowledge and the complexity of IoT Cloud Systems, the developer might introduce or might not be able to detect various types of uncertainties, which strongly influence the application. In this talk we aim at detecting such uncertainties and recommend software design to deal with such uncertainties as early as possible. We model and evaluate potential uncertainties on design artifacts representing structural and/or behavioral information about the system under study. We propose a rule-based Uncertainty Modeling and Evaluation methodology (UME) and tool (T4UME) to help users in detecting potential uncertainties on design artifacts and to decide whether or not refactoring strategies should be applied to uncertain system design artifacts. In particular, our framework deals with uncertainty as a crosscutting, multidisciplinary concept by providing proper extension and customisation mechanism to suitably tailor its adoption to different domains.