In probabilistic seismic hazard assessment, the development of the seismic source characterization, especially the geometry of the seismic source models (SSMs), is controversial because it often relies on expert judgment with different interpretations of the available data from seismology, tectonics, and geology. Based on the same input datasets, different teams of experts may derive different SSMs. In this context, the verification of the models through the comparison against a set of observations is a crucial step. We present a statistical tool to compare the SSMs with the observed seismicity and rank these SSMs based on their ability to replicate the past seismicity. We simulate many synthetic catalogues derived from candidate SSMs and compare them with the observed catalogue of mainshocks using the Metropolis-Hastings Algorithm to select those that fit the observed catalogue. The candidate SSMs are then expressed by a probability density function (pdf) using the set of synthetic catalogues accepted by the Metropolis-Hastings Algorithm and the Bayesian inference. To help practitioners in earthquake and civil engineering understand how this tool works in practice, the proposed approach is applied to a proposed new nuclear site in the United Kingdom, Wylfa Newydd.