A/B Testing Basics: What You Need to Know

June 13, 2012
118 Views

Every person has varied interests and different preferences. What looks nice or acceptable to someone might not be as presentable to others. This applies to almost everything perceivable in the world, from clothes and shoes to website layouts and designs.

 

Every person has varied interests and different preferences. What looks nice or acceptable to someone might not be as presentable to others. This applies to almost everything perceivable in the world, from clothes and shoes to website layouts and designs.

 

The website design process itself is usually long and arduous. Aside from having to conform with most design guidelines and recommendations, web designers should also take into consideration their target audience’s personality and preferences. They can assess this through surveys and by doing research on their intended users, but one thing that could help them with this process is testing.

 

Testing Methods: A/B Testing and Multivariate Testing


There are a number of testing methods around, but two of the most common ones in use nowadays are A/B testing and multivariate testing. Both are similar in the sense that they allow for the comparison of two versions of different websites, each with certain elements tweaked or modified. However, only one element is to be changed in A/B testing so that the reason for the varied reaction from the target users can easily be identified.

 

A/B testing is generally recommended for websites that contain a smaller number of elements. This is because conducting a series of these tests will require a longer amount of time. Following this, multivariate testing is best suited for sites that are complex in nature or contain a larger number of components.

 

How Does A/B Testing Work?


As mentioned earlier, two options are presented to the user in an A/B test. For websites, these two website versions are served simultaneously and randomly to site visitors. For email campaigns, marketers can simply send out two different mailings to separate groups in their lists to carry out the test.

 

A criterion for determining which version is the better one is usually set before the test is conducted. For example, if you’re primarily concerned with conversions, then the variation you should choose is the one that resulted in the most leads or clicks. If you place a higher importance on the actual sales made, then you should go with the version that gave you more sales.

 

A/B Testing on Websites


To conduct an A/B test on websites, you must first perform a general assessment of your web pages. Note down any elements that could still be improved, and re-arrange your list to indicate which ones should be prioritized. Examples of where you could possibly run the test on are the banner, image placement, site layout, navigation menu, site theme, and so on.

 

When you’re ready, prepare the versions that you will be using for your test, then set up your server to display the two versions.

 

A/B Testing for Email Campaigns


Running these tests for online marketing email campaigns can be done in a similar fashion. Bring up copies of past emails you sent out to your list and take note of which elements you’d like to change. Some examples could be your writing tone or style, use of images, layout, and calls to action. It’s easy to set up the email manager program you’re using to send out a split mailing of the email to your list.

 

Tools and Software for A/B Testing


The wide availability of apps and software makes it easy to run your own A/B tests. One tool that’s available for free online is the Google Web Optimizer, which will guide you through the entire testing process so you can run them yourself with ease. If you prefer, you can also opt to hire a third-party firm to run the tests for you so all you need to do is discuss the test specifics and wait for the results to roll in. To find out more, check out Maxymiser’s A/B testing guide