To understand the opportunities and challenges of large shares of solar photovoltaics (PV) in our electricity mix, various large-scale studies of PV potential on building roofs have been conducted in recent years. The use of different datasets, methods and spatio-temporal resolutions leads to widely varying results, making a comparison across different studies intrinsically difficult. In this work, six studies carried out in Switzerland are compared in a quantitative way, in order to understand how different methods impact the potential estimates. We observe a strong trend towards increasing spatial and temporal resolutions, using larger and more accurate datasets for the analysis. While the earliest study relies on rules of thumb, later studies use data-driven estimations, enabled by the use of Machine Learning, scalable algorithms and powerful computational engines. Our analysis shows that the largest differences are caused by the source of the solar radiation input data, the computation of shading effects on rooftops and the estimation of available roof area for PV panel installation. The latter is the most uncertain parameter in the presented studies and offers opportunities for future work.