While clustering is a powerful methodology used for grouping objects into families, it is hard to conceive a natural object grouping method without considering the context of a particular application. In high-dimensional problems with large volumes and rapidly evolving part flows, many clustering methods have traditionally been used to form part families considering similarities. In this paper, a two-phase clustering method is developed to optimize the grouping process under technological constraints, in a bid to improve resource planning. The first phase consists in forming part families using the Agglomerative Hierarchical Clustering approach, considering multidimensional parametrization of the part, while in the second phase, the optimal number of clusters is determined using ELECTRE III, which serves to handle uncertainty. Based on a real case study in the electronics industry, an improved production planning solution is proposed to validate the method’s efficiency. The solution was compared to the in-use method to highlight its added value.