A regret order based clustering heuristic is proposed to create capacitated clusters for spatial datasets with equal weighting. This heuristic can be used in a vast number of applications, including the Capacitated P-Median problem (CPMP) and Capacitated Facility Location Problem (CFLP). The heuristic is scalable and therefore also useful to cluster large instances. The heuristic builds on an initial uncapacitated clustering solutions the can be generated by various clustering methods and has proven to provide good quality end solutions independent of the quality of the initial solution. Different types of regret proximities are tested for the Capacitated P-Median problem (CPMP) in particular. The results along with visual plots on a map of geometrical data are shown to illustrate the impact of the different proximity rules.