Energy communities (ECs) aggregate users within proximity, which have diverse assets and consumption/generation power profiles. Such a variety of user arrangements significantly influences the benefits expected from the ECs. From a vast pool of EC configurations, this paper investigates their composition regarding users' profiles impact on collective benefits. To that end, clustering is performed for i) different features to characterize an EC and for ii) the performance metrics of the EC once managed. The paper discusses how the community setup impacts its performance, which enables identifying the most relevant features. Hence, 1000 ECs are formed from 10 users. Additionally, two study cases are tested, one with 100% users with PV and battery (i.e., 100% prosumers) and the second with 50% prosumers. The results suggest that in terms of investment, the photovoltaic installed capacity is a more significant asset than storage capacity.