abstract: We will explain the techniques used to compute rigorously the invariant measure and rigorously estimate observables for general dynamical systems with noise. We will illustrate the interplay of Wasserstein, $L1$ and variation norms on the space of signed measures, and show how it can be exploited to obtain a surprisingly effective estimation of the invariant measure in the $L1$ norm. We conclude showing how this allows to prove rigorously the noise-induced order phenomenon for a model of the Belousov-Zhabotinsky reaction, that had been discovered with numerical simulations by Matsumoto-Tsuda in 1983.