We apply the principles of Bayesian statistics to the main probes of cosmology, in order to refine our knowledge of the Standard Model and possibly extend it. Notably, we investigate the basic elements of the model in detail in order to reinforce this basic foundation of the field, and lay down a systematic way of obtaining model-independent constraints on parts of the Standard Model. We further try to constrain some of the unknown properties of Dark Matter, namely its decay or annihilation rates, to help reducing the range of possibilities for model builders. By using recent cosmological probes and making as little assumptions as possible, we are able to meaningfully constrain these properties in the prospect of narrowing down a particle physics search. Eventually, we show how future experiments will be able to put strong bounds on the neutrino total mass, as long as the theoretical uncertainty is handled carefully. Despite being cautiously pessimistic, we prove how EUCLID will be able to detect even the lowest possible allowed neutrino mass, by simply using properly the linear scales. We also show the target precision for the theoretical prediction in order to make full use of the forthcoming wealth of data at mildly non-linear scales.