This work covers general multistage binary separations with application examples in cooling crystallization, evaporative crystallization and organic solvent nanofiltration. Deterministic global optimization is applied to identify optimal configurations of multistage separation networks and study their sensitivity to parameter values. Superstructure optimization is conducted for countercurrent cascades and also for general superstructures to identify new multistage configurations. Results show substantially reduced separation effort for alternative configurations in large parameter regions, specifically in regions where optimal countercurrent cascades have a low number of stages. General, simple design guidelines for multistage separation with a low number of stages are derived from rigorous global optimization by comparing results for different processes.