Background Building energy efficiency assessment plays a crucial role in sustainable development efforts, and current methods regularly use overly complex models where input data cannot always be known for existing buildings. In this study, a predictive algorithm is used to automatically assess the energy efficiency indicators of buildings based on the Swiss Building Energy Certification (CECB) requirements and reduce the time spent in the decision-making and early phases of a renovation project. The algorithm aims to provide accurate estimations of heating needs, final energy consumption, and CO2 emissions, considering both renovated and non-renovated elements. Purpose The primary objective is to evaluate the performance of the algorithm in predicting building energy efficiency indicators and to identify its strengths and weaknesses. Additionally, the study aims to propose recommendations for enhancing the algorithm's accuracy and applicability. Methods Detailed comparisons are presented between the algorithm's predictions and CECB dataset results for various buildings constructed before and after 1970. Analyses are also conducted on individual buildings to examine discrepancies and factors influencing energy efficiency estimations. Furthermore, statistical analyses are performed to assess the algorithm's performance in terms of letter rating consistency with the CECB dataset. Findings and interpretation The findings reveal that the algorithm demonstrates promising performance for buildings constructed after 1970, providing more accurate estimations of energy efficiency indicators. However, challenges arise with older buildings and those with unique typological compositions, leading to overestimations or underestimations of energy needs, final energy consumption, and CO2 emissions. The incorporation of renovated elements in the model offers potential for more precise predictions but introduces complexities. Statistical analyses indicate that the algorithm's outcomes align closely with CECB ratings in many cases, although improvements are needed to address discrepancies. Conclusion and recommendations In conclusion, this study underscores the importance of refining the algorithm to accurately represent diverse building typologies and renovation scenarios. Recommendations for improvement include incorporating variable typological compositions, introducing simple user interactions, refining the typological window-to-wall ratio, and leveraging more reliable data sources. By addressing these areas, the algorithm can become a valuable tool for experts to rapidly assess any building’s performance before entering advanced stages of the project, without requiring overly complex models.