Often you want to experiment with two alternatives of user experiences for the same problem, sometimes you want to improve the experience but are not 100% sure so you like to expose new variants to only a few users and get statistics. A/B testing helps you identify which option is more attractive. Having more than one option is also possible but that increases the complexity.

Examples could be what you emphasize in your content or which colour you use for your buttons. You need a good analytics system in place to measure the success of each option which helps you decide.

• Read More on Wikipedia