With landing page split tests, you can test up to 5 different landing pages to see which one is most effective to help reach your goals.
To create a split test, navigate to the Sites tab. You can create a new landing page or use an existing one by clicking on the name. If you click on a landing page name, you'll get to the overview page. Now click on the orange button that says Create split test. In the next screen, click Add your first version. In the pop-up, you can choose to create a landing page from scratch, copy from the original or copy from another landing page. We're going for the original option in this video.
You'll then get to the editor where you can make the change you want. For a split test, it's best to test 1 thing at a time. For example, change the button text and click Save. By only having 1 difference between the 2 landing pages, you can guarantee that the button text is responsible for any statistical changes you'll see between the 2 pages.
After you click Save, you will see your new version underneath the original one. It's good to know that landing page testing is based on unique visitors traffic, not on page views. Underneath your landing pages you can decide the distribution of the traffic. It's best to choose 50% and 50%. You can also add more versions to test—with a total of 5 different versions.
Select the ON slider to activate your split test. Remember that split tests start immediately and can't be stopped until a winner is selected. You can only start an A/B split test for landing pages once—so it's good to think things through before you start one.
When you want to turn your split test OFF, you'll need to select a winner. You can base the winner on the unique subscribers, views and conversion rate. The winner depends on the goal you set initially. After you've picked the winner you can't set up another split test or choose another winner. If you'd like to do this, you'll need to duplicate the landing page. The reason for this is that your data stays clean. For each test you start with new statistics, so you can accurately measure the results.