Big data are already starting to transform business, government, science and everyday life. The increased capacity and reduction of costs in sensing and connecting devices are indeed providing a deluge of data which will impact at individual and global level. Personalized medicine, border security, traffic control, custom manufacturing and delivery, are just some examples of fields which will be drastically impacted by the proliferation of data. Data-driven services and products can provide a unprecedented opportunity for boosting a new generation of jobs. But data are not information or knowledge. The analysis and interpretation of evidence require indeed the understanding of the essential variables, the involved scales, the quality and interoperability of different sets of data, the robustness and accuracy of the results provided through their analysis. The complexity of the involved systems can in fact induce misleading interpretations. Data sampling, meant not only as numbers, and mathematical models have therefore to be adequately and strategically matched to provide accurate evaluations to support decisions.