Few applications of standard bootstrap techniques exist in the literature in clustering problem settings. This paper presents an approach to enhance clustering methods using proper Bayesian bootstrap: in the proposal, prior knowledge about the group densities is introduced, a gaussian mixture model is taken as the prior for the distribution of data. Our approach is organized in two steps. Efron bootstrap is used to estimate the parameters of the prior. Proper Bayesian bootstrap is used to resample from a mixture of the prior and empirical distribution of data. Novelty elements and robustness of the proposal are presented as empirical evidence on the Iris dataset.