A/B testing is the practice of testing a single variable of an ad and keeping the rest constant. For example, we can test ad A against ad B, the only difference being the headlines. In this example, the sole purpose is to find out which headline brings in better results for the business.
Multivariate testing is an optimization strategy that allows testing multiple variants of ad elements to determine the most effective combination for a single ad. Rather than just changing one ad element, like the headline, you can change a group of different elements to see which combinations perform better with your target users. Sometimes there are just too many elements of an ad to be tested, so multivariate testing can cut short the time needed to find a winner.