Building regulations, scarcity of energy, and climate change have forced designers to find energy-efficient design alternatives for the buildings. Current regulations focus solely on the total energy requirement of the building without considering the fact that the energy performance varies greatly across different units of the building, which, in turn, causes discomfort among the occupants. Conventional optimization approaches created based on these regulations, therefore, miss the capability to cope with this issue. Resolving the problem of varying thermal performance within the units requires the introduction of unit-based optimization approaches. This study elaborates on revealing the inadequacy of the conventional optimization approach and proposes two alternative approaches that take the issue into account. Within this context, the thermal design a typical five-story residential building with six apartment units on each floor was optimized according to the conventional optimization approach. A simulation-based optimization system consisting of a Distributed Evolutionary Algorithms in Python (DEAP) optimization tool and Energy Plus was employed. The differences in the energy performances of different units were observed for three different climate conditions. Afterwards, two different approaches having the objectives of optimizing the overall building performance and balancing the variance within units were proposed: (i) single-phase multi-objective optimization and (ii) multi-phase single-objective optimization. The outcomes of the study demonstrated that the multi-phase single-objective optimization provided better results.