For example, you want to see if shifting a “Buy now” button to the sidebar of a page instead of keeping it on the top will improve its click-through rate. The current design is called the “control” version (version A). Version B is called the “challenger.” Then you can start testing these two versions by showing them to a predefined number of site visitors to see which version is clicked more. It’s good to divide your sample groups equally and randomly.
Usually, A/B testing is used to test landing pages, welcome emails, and ads. It is better to focus on things that are most likely to have a big impact: search modals, call-to-action buttons, headlines, images, product descriptions, navigation, social sharing buttons etc.
The terms ‘A/B testing’ and ‘split testing’ are often used in the same meaning. But in fact, they are two different types of tests – A/B testing compares two versions by changing one element and split testing compares two distinct designs.