Four-wire low-voltage microgrids supply one-phase consumers with electricity, responding to a continuously changing demand. For addressing climate change concerns, national governments have implemented incentive schemes for residential consumers, encouraging the installation of home PV panels for covering self-consumption needs. In the absence of adequate storage capacities, the surplus is sold back by these entities, called prosumers, to the grid operator or, in local markets, to other consumers. While these initiatives encourage the proliferation of green energy resources, and ample research is dedicated to local market designs for prosumer–consumer trading, the main concern of distribution network operators is the influence of power flows generated by prosumers’ surplus injection on the operating states of microgrids. The change in power flow amount and direction can greatly influence the economic and technical operating conditions of radial grids. This paper proposes a metaheuristic algorithm for prosumer surplus management that optimizes the power surplus injections using the automated control of three-phase inverters, with the aim of reducing the active power losses over a typical day of operation. A case study was performed on two real distribution networks with distinct layouts and load profiles, and the algorithm resulted efficient in both scenarios. By optimally distributing the prosumer generation surplus on the three phases of the network, significant loss reductions were obtained, with the best results when the generated power was injected in an unbalanced, three-phase flow.