Microgrid operation scheduling is a widely studied topic, but the performances of the solution algorithms proposed in the literature are rarely discussed and, in practice, scheduling is often only loosely optimized. In this paper, we present a practical context of coordination between a microgrid and an utility grid which concerns hundreds of microgrids worldwide. We adapt three distinct optimization strategies-mixed integer quadratic programming, simulation-based genetic algorithm and expert-based heuristic-and empirically compare their strengths. Our results emphasize the benefit of seeking for optimality, while the inaccuracies of the system dynamic model are compensated by periodic recomputation.