Island-based metaheuristics have gained significant attention in the field of optimization due to their ability to maintain population diversity and avoid premature convergence. A critical component of these algorithms is the migration strategy, which determines how individuals are exchanged between islands. This paper investigates the impact of different migration strategies on the performance of island-based metaheuristics, with a particular focus on the number of migrated individuals and the criteria for their selection. We propose several strategies for selecting individuals for migration, including random selection, fitness-based, diversity-based and hybrid approaches, and evaluate their effectiveness on a set of TSP (Traveling Salesman Problem) and BBOB (Black-box Optimization Benchmarking) problems. Our results demonstrate that the choice of migration strategy significantly affects the algorithm’s performance. Specifically, selecting individuals based not only on fitness but also on their potential to increase diversity leads to better outcomes.