AB Testing: when and how to properly conduct an AB test
You wouldn’t buy a computer without determining whether it meets your needs, so why would you redesign your website without an AB test? AB testing has become a key component of any successful website redesign, particularly so for e-Commerce enabled websites, or so you may have heard. But many may be left wondering, “What should I test?” and “How do I conduct an AB test properly?”
To answer these questions, let’s start with the basic idea behind AB testing. An AB test works by displaying two different versions of a web page to your viewers at random, and analyzes the activity on each version to gauge which is likely to produce higher customer conversion rates. This is one of the main ways designers determine best practices in website design and layout, and one of the main methods for copywriters to determine what words and phrases are most likely to appeal to a particular demographic. But before you jump right in and start testing random changes, it’s important that you consider the following steps to get accurate, useful results:
1. Create a hypothesis to test
A good hypothesis on which to conduct an AB test would be something like, “Removing the phone number field from the form to download our free report is likely to increase form completion rate, and subsequently, sales of the product or service our free report explains, because customers will feel more comfortable filling out the form if they know they will not receive unsolicited calls.”
2. Create the two page versions and testing parameters
One of the two page versions you test should be the page you currently have online. This acts as your control, and will provide you with analytics data against which you can compare the new version. To set up the test parameters, you will need an AB testing program (or a marketing consultant to set up and run the test for you) and analytics code to install onto each version of the page.
3. Analyze the results and implement the winning solution
Unless you are familiar with web analytics and AB testing, you will probably want to leave analyzing the data to a marketing consultant as well. If the data supports your hypothesis and your new page outperforms your existing one, make the change permanent online. If the new page shows no higher conversion rate, or worse, a decline in conversions, it’s time to go back to step 1 and reformulate your hypothesis. Remember, a negative test result doesn’t mean your page is perfect and you should do nothing ; there is always room for improvement.
4. Run multiple variations of the winning solution
Returning to the example of the download form, if the data supports your hypothesis that removing the phone number field will result in higher completion rates, you may want to experiment with the position of your new form to find the ideal placement within your site’s visual hierarchy, such as placing it on the left-hand side of the page versus the right-hand side. This is called a real estate test. Or you may wish to try removing words or graphics you feel are extraneous to the form, which is commonly referred to as an inclusion/exclusion test.
If you decide to hire a consultant to conduct your AB testing, Niche Marketing PR is ready and willing to assist you with setting up, launching, and analyzing each AB test you desire.