Bayesian statistics relies on the fact that the people can use previous experience to be able to accurately forecast or estimate the future events or derive conclusions on statistical population. It differs from frequentist inference in that it takes into account the results of previous observations and uses those to moderate the calculations for current calculation.
The Bayesian statistics (or theorem) in CRO is mostly applied for the purposes of A/B testing. In contrast to frequentist statistics testing, Bayesian testing does not require a specific sample size or statistical significance. Instead, the performance of experiment is measured constantly and the test provides the information how likely the variation is to beat the baseline.