The knowledge of reservoir fluids phase behavior has always played an important role in oilfield development planning, reserves evaluation and screening of the potential for enhanced oil recovery. Nowadays operators aim more and more at fast-track development of discovered resources, therefore, any anticipation of thermodynamic properties is a business challenge: looking for "PVT-analogues" is the solution proposed in this paper. What adversely impacts massive scouting of PVT data usually is the limit of a small amount of readily available information, also due to the intrinsic complexity of the datasets and of the variety of output formats produced by different laboratories all over the world and over the years. In Eni a new tool for data mining based on the reorganization and thorough digitalization of the PVT archive is in advanced development. Standardization of the laboratory outcomes by templates, automatic loading into a corporate repository, in-house development of software tools for quality control, data mining and advanced statistical analyses, easy access through a properly designed interface: each of these steps is integrated in an upgraded data-driven approach to fluid properties prediction allowing an earlier understanding of the reservoir fluid system.