With the increasing penetration of intermittent renewable energy and the electrification of various usages, demand-side flexibility has become essential to maintain the balance of power grids. Buildings, accounting for 30% of the world’s total final energy consumption, can play a key role in this requirement by modulating their consumption. Moreover, metering data has become more accessible and of better quality worldwide, opening up opportunities to develop flexibility services based on robust and reliable data analysis in non-residential buildings. This study aims to develop a data analysis methodology to quantify the electrical flexibility potential of a building and to determine the influence of data type in the quantification process. To achieve this, three different methods based on three different types of data are tested to quantify this potential and to identify the added value of more detailed data. The analysis was performed on a dataset of electricity demand measurements collected over a period of 11 months in university buildings in France. The results show that the third method, based on time series of electricity consumption data at the sub-meter level, provides the best accuracy with a CVRMSE of 22% and a NMBE of −4%.