A next generation of models is needed to guarantee energy performances of buildings and systems in practice. Current energy performance models use assumptions on climate, user behavior, characteristics of building and systems that may deviate from reality. Improvement of the accuracy of these models and assumptions is necessary to better predict the real energy performance of buildings and systems. Joining calculation models with data, using dynamic parameter identification (DPI), is a promising way for a better prediction of energy use in practice. Three example cases for identifying deviating building and system properties are discussed.