If you aren’t conducting tests on your website, then you need to be. If your website isn’t performing at its optimum level, then you are leaking revenue. But I’m not writing to convince you that you should be testing. I’m here to explain what makes a test valid.
The most important component of running a test on your website is defining what you are testing. You must clearly define your test with a question (i.e. Which layout design will have the lowest bounce rate? or Which call-to-action will drive the most conversions?). If you don’t define your test, then the results will be meaningless and you will have wasted your time.
After you’ve clearly defined your test, you will set up each version (we call them treatments) of your test. Then, you need to define your goal, or the desired level of improvement that you want to see.
Statistical formulas are used to decide how long a test must run to produce valid results, and whether the results have indicated a winning version or treatment. Every test will not have a winning treatment. This is a key principle of testing that you must remember: a test does not have to have a winning treatment in order to be a valid test.
It is important to record the results for each test, even if there is not a winning treatment, along with all of the insights you gain from each test. Sometimes a test without a winning treatment will give you more insights than one with a winning treatment.
As you continue to test, you should begin to build a testing history that will lead to valuable insights and ideas for new tests. So start your testing history today and see how much you can improve your website results.