A/B testing, also known as split testing, is a way to compare variations of the size, color and arrangement of your web pages’ elements in order to let you know which version provides optimum conversion. Split testing can let you know which changes increase conversion rates, unique visits, lead captures, signups and other calls-to-action.
A/B testing can reveal the truth about your design assumptions and alert you to interpretations you didn’t anticipate.
Where do you start?
When you perform A/B tests on your site you will want to keep your goals clear and in order. Ask yourself, “What do I want my visitors to do when they visit my site?” Once you’ve answered that question, consider all of the basic page elements you have at your disposal that will facilitate that goal: your logo, action buttons, heading text, input forms, informational text, photography, and navigation items. Create a design that arranges these elements according to what you think will best encourage fulfillment of your goals. Put your best foot forward and start your tests. If you use a platform like Google Analytics Content Experiments to perform your A/B tests and have steady traffic to your site, the results should start to favor A or B within a couple of days.
What do I look for with my test?
A/B testing is not simply about identifying which page performs best. It is about identifying why a particular page performs the way it does. The former tells you what’s popular. The latter tells you what your visitor wants and needs. That’s the most important part. Otherwise, you’re still guessing why your visitors took one path over another.
It is important to remember that when you test multiple changes simultaneously it can be difficult to determine why exactly a particular version was successful or unsuccessful. For instance, if you make a version where you move the logo and change the text in the same variation, then you have no way of knowing if the key factor was the position of the logo or the new text. The most valuable tests are ones that give you information on what your users prefer not just which version wins. One of the great things about optimizing your web site is that you can’t lose if you do your tests methodically. Each new test will give you the answer you need and eventually you will end up with a version that is better than the one before.
What else can I do?
Consider using a heat map tool like CrazyEgg to enhance your understanding of which parts of your site visitors expect to be clickable. Traditional analytics software like Google Analytics can tell you which pages are getting attention relative to other pages on the site; but they can’t tell you how far down your visitors scroll, which images and headlines they expect to be clickable or whether they ever saw your call-to-action button.
Optimizely is a popular tool used for A/B testing. It is a code generator that allows you to create variations of your website by inserting a snippet of code into the head of your pages to be tested. The code is generated through an online editor that lets you resize, move, hide, edit your HTML and more. Unbounce is another A/B testing tool, but, unlike Optimizely, Unbounce includes landing page templates, lead capture forms, email lead notifications, Facebook “like” conversion tracking and even mobile landing page templates.
A/B testing can be a very useful process to help you optimize elements on your site that will lead to higher conversion. Sometimes a particular design choice versus its outcome is not intuitive. You don’t always have to take the recommendation of an A/B test if you feel it may compromise your brand or messaging. Use A/B tests to help you validate design assumptions and produce the best business result for your web site.