What is Split Testing?
Split-testing (Also known as A/B testing) is a statistical tool for measuring the effects of slight variations to a page on your website. It is great for driving continuous improvement, and raising your conversion rate. You basically present two slightly different versions of your web page to your visitors, and track the conversion results of each version.
Why Should I Split Test?
It has been scientifically proven that certain elements of your web page can have drastically different effects on visitors based on size, color, placement, and a host of other variables. The only way to know which elements work best for your visitors, is to test them against one another. The main reason you split-test is to increase your sales, leads, or any other type of conversion.
What Should I Test?
You can test any element of your page you wish. Some common things to test are:
- Headings and Sub-headings
- Opening Paragraphs
- Call To Action buttons
- Location of different elements (such as the shopping cart)
How Do I Split Test?
The first thing you need is two slightly different versions of your web page. Be sure to only change one element at a time or you will have trouble determining the significance of your results.
Next, you will need some type of software application to deliver the different versions of your page to your visitors. Google Website Optimizer is a great one that is free. It delivers alternating versions of your web page to visitors (Visitor 1 gets A, Visitor 2 gets B, Visitor 3 gets A, etc), tracks results on your conversion rate, and delivers the results to you. The advantage it has over other platforms is that it allows you to use your existing URL, which means you can track traffic obtained from all sources. Using the optimizer should not have any negative effects on your rankings or cause duplicate content issues.
Who Should Do The Testing?
It is best to have your web designer set-up and run the tests. They should start immediately after launch and continue until you feel that your web page is the absolute best it can be, or the best you need it to be. It is possible to use a third party down the road, but much more difficult. You will also give valuable experience to your designer on what elements work best that will be useful on future projects.
How Do I Know If It Worked?
Measuring your results requires determining whether or not they are statistically significant. Simply put, the difference between the two pages is significant if the numerical difference between the two is greater than the square root of the sum of the two results.
Y-X>√X+Y, where Y>X
Now that your brain has exploded, let me give you a concrete example.
- Page A resulted in 50 conversions
- Page B resulted in 70 conversions
- The Sum of the results of Page A and Page B is 120
- The Difference between Page A and Page B is 20
- The square root of 120 is 11, rounded up.
- 20 is greater than 11, so the difference had statistical significance
Now you have some idea of how split-testing can improve your conversion rates. Use it often, and you may be amazed at the results you receive.