Energy storage systems are becoming one of the most relevant technologies to effectively support renewable energy source (RES) deployment at large. The present work proposes a detailed ageing and energy analysis based on a data-driven empirical approach of a real utility-scale grid-connected lithium-ion battery energy storage system (LIBESS) for providing power grid services. The system under investigation is an operative utility-scale LIBESS integrated with a multi-MW PV plant and connected to the medium voltage network, located in Southern-Italy. The ageing and energy analysis has been performed using data measured by the supervisory control and data acquisition (SCADA) system directly connected with the LIBESS. This large amount of data is collected in a cloud database and then elaborated using proper software tools. This experimental campaign applied on a commercial LIBESS covers the impact of degradation mechanisms, such as cycle and calendar ageing, the battery and global system efficiency as well as the role of auxiliaries' power consumption under normal and power grid services operations. Thanks to the low complexity of the proposed data-driven model, it could be easily replicated in any other LIBESS facility, both operative in real-world framework and in laboratory context. The state-of-health (SOH) estimated by the degradation model implemented in this work after 3 years of operations and after 356 full-cycles equivalents is equal to 95.88 %, with an average capacity loss compared to the nominal capacity of about 1.37 % per year. Energy analysis revealed an average system global efficiency of 85 % for operations close to the nominal power that decreases to 65 % for operations at low power rates. While considering the grid network services operations, the power grid applications investigated are primary frequency regulation, secondary voltage regulation and PV unbalances reduction. Our results show that primary frequency regulation is the most efficient service in terms of ageing and energy performances. It has been evaluated a capacity loss of 0.03 kWh per each full-cycle equivalent performed during primary frequency regulation. Differently, secondary voltage regulation and PV unbalances reduction present a capacity loss of 0.1 kWh and of 0.05 kWh per each full-cycle equivalent, respectively. At the same time, the global efficiency of primary regulation (around 83 %) is remarkably higher compared to secondary voltage regulation (about 47 %) and to PV unbalances reduction (about 55 %). Results support the relevance to identify the most impacting key performance indicators for effective exploitation of LIBESS in power system and network services applications.