A round-robin test on a BIPV-module has been performed within the framework of the IEA PVPS Task15. Identical modules from the same series have been measured indoors under STC-conditions to be equal in rated power within a largest difference between lowest and highest performing module of 0.3%. Nine sites spread over the world – with a clear focus on Europe – measured one BIPV-module from this series in a well-described mock-up that mimics a ventilated cold façade. Due to storm damage only seven sites could measure a long enough period to be of value for the comparison. The longest measured period has been around 2 years. However, for each site the best one-year period with hardly any missing datapoints has been selected. Data filtering was needed to account for various measurement failure and errors that are typically occurring in outdoor performance measurements. Finally, datasets of around 40k datapoints per site (of 5-minute resampled time resolution) were analyzed by the IEC 61853-matrix approach. The matrix from this standard was extended with other parameters of interest. Because all sites have a different climate, the matrix is populated in different areas. Therefore, the PR-matrix was found best suited to use as a comparative tool with a threshold for each bin of the matrix set at minimal 0.5% contribution to full-year specific yield. The measurement accuracy in PR for each bin was found to be about 2%, of which the irradiance sensor (either calibrated reference cell or secondary standard pyranometer) is the most contributing. The finally composed bin-wised averaged PR-matrix shows large areas of small difference between sites that are about equal to the measurement accuracy. But there are also areas of the matrix in which distinct features of a site can be observed. Most of these features could either be explained, or are most likely to be originating from an Angle-of-Incidence effect or spectral effect, which are outside the scope of this paper. In conclusion, the authors like to propose that one can quantify the energy performance of a (BI)PV-module in a well-defined mock-up, within a full measurement system accuracy of around +/-2% (with the irradiance sensor as bottleneck), independent of the local climate, following the measurement and analysis protocol of our round-robin as described in this paper.