Thanks to its intrinsic linear properties, the application of voltage sensitivity coefficients theory for the development of fast and effective low voltage (LV) distribution grid control algorithms knows a strong interest among the scientific community. In particular, model-less methods for computation of sensitivity coefficients received lots of attention, as the accurate grid model (e.g., line parameters) is not always available in LV grids. In this paper, a model-less/measurement-based method for computation of voltage sensitivity coefficients in a real distribution grid under known operation scenario is evaluated and experimentally validated. The model-less method for computation of sensitivity coefficients has been implemented in a decentralized measurement and controldevice called GridEye. GridEye is a digital grid optimization tool to design, operate, monitor, analyze, and automate any power distribution grid. The tests are performed in the “ReIne” laboratory, a new flexible structure, with nominal power up to 100kVA, able to emulate different distribution network topologies, such as radial, meshed or partly meshed network.