In the European Union, while 95% of the existing infrastructure will be in use by 2050, only 25% is energy efficient. Although building-energy retrofits are needed to improve energy efficiency of the European Building Stock, their current efficiency is limited due to mismatch between the predictions and measured values of energy use. To address this challenge, measurements have been combined with building-energy models during the retrofit process. It is valuable to identify a suitable model capable of describing building-energy performance in a manner that is consistent with measured data. Previous approaches validated models through residual minimization. Residual minimization has had the shortcoming of not accounting for uncertainty and performing poorly when extrapolating. Another researched approach was Bayesian Model Update (BMU). While BMU accounted for uncertainty, zero-mean Gaussian distributions were often assumed, which is not realistic in the built-environment context. Efforts to adapt BMU to include other distributions have resulted in complex formulations, limiting its practical application. To address these limitations, a variation of BMU named Error Domain Model Falsification (EDMF) was developed in the structural health monitoring field and has proven useful in guiding structural retrofit decisions of existing bridges. The present paper adapts the EDMF method to the building energy retrofit context for supporting decision making related to building energy retrofits. The strengths and limitations of EDMF are discussed. The present study shows how EDMF reduces the uncertainty of values of building-energy-modelling parameters, and thus, reducing the performance gap.