Menufy.com, a restaurant order & delivery service, ran a multiple change A/B test on their checkout page (with our help). I really like this experiment because it's effect is rather small and yet tightly defined in terms of its statistical significance. Sometimes when we make UI improvements we might feel that our work will have a great impact, and this test brings us down a little closer to reality. Here are the changes which were tested and in return inspired our evidence-based patterns.
The 4 Changes
Here are the 4 things that were changed (interpreted by me).