A comparison of two versions of the same page, where traffic is split randomly. It measures which version leads to the desired action more often. In B2B, A/B tests often fail due to sample sizes that are too small: without enough visitors, the result is just a matter of chance.