For IFRS9, the computation logic of the ECL (Expected Credit Loss) is not particularly diffcult, however, the analysis of the variation and change of credit worthiness can be very complex, require twelve month of historical data and the role of the risk analyst is key to assess if a given contract should be in stage 1, 2 or 3, according to the IFRS9 regulation. This analysis require a lot of attributes (days passed dues, rating change,…) and this is the reason why it is so important to be able to ddrill down to the most granular level on the dataset. Working on pre-aggregated data on IFRS9 is not an option anymore.