The objective of the SHOWE-IT project is to demonstrate, under real conditions, how advanced ICT components and systems can enable services that help reduce energy and water consumption in social housing across Europe. To achieve this, the project takes a demand-driven approach, prioritizing as starting point an affordable investment per dwelling, and putting in place an integrated and easy replicable ICT-based service. The project expects to achieve an overall energy and water consumption reduction of 20 %. In SHOWE-IT, the main scientific bottleneck is due to the lack of comparative data to estimate the savings, once the technology is put in place. Considering the context of the project, the option that fits better is the CGPG method (Control Group/Pilot Group). The CGPG consists of establishing two groups with the same kind of profile in each location. One group has intervention (pilot group) and the other (control group) have not. The savings are calculated by an indirect comparison between those two groups. A fully automated kriging metamodel is used to estimate the consumptions that would probably correspond to the Pilot Group if there were no ICT treatment. The challenge here is to select automatically the main socio economic variables that better explain these consumptions on the one hand, and increase the model accuracy to make the estimated savings reliable on the other hand. The first results on electricity consumption show an average savings lower than 10% and a confidence interval at 9%. The quantity of data (only two months) explains those results. In fact, we expect the uncertainty to decrease and we hope the estimate savings to increase on duration. However, to optimize the methodology, ongoing works are developed on two points: 1:) The robustness of the methodology since the data contains aberrant observations. 2:) The optimization of the variables selection phase by combining the automated criteria to field expert's knowledge.