In optimizing your website, sometimes you want to test something simple, like a red button vs. a blue button, or one kind of product image against another. You should be testing those sorts of things — often.

Other times you want to test a complete redesign. Those are also worthwhile, but it’s harder to know how to interpret the results.

If you’re only changing one thing on a page, any change in response can be attributed to that change. When you’re changing a whole lot of things, you don’t know which element caused the change in response.

Let’s say you test a new design against your current design and the new design loses. You might be tempted to give up on the new design and stick with what you have. But that may be a mistake. There may be one small thing on the new design that’s causing it to lose in the test.

Before you give up entirely, try tweaking the new design a few times. You may find that a small change makes all the difference.