What is A/B Testing?
CREDIT: CoraMax | Shutterstock
In the world of marketing and Web design, A/B testing is used to better understand website visitor behaviors with the intended goal of increasing conversion rates. Conversion rates indicate the point at which a website visitor finally makes the decision to purchase a product or fill out a form. A/B testing, also known as split testing, is the practice through which marketers understand which design choices to make on a website to facilitate that buying behavior.
Despite the successes found in A/B testing, few companies make use of this strategy as opposed to search engine optimization (SEO), web analytics and simple web page usability. Part of the reason behind this is simply that few companies know about it or how to use it.
A versus B
The use of A/B is meant to indicate different versions of the experiment. For example, version A is used as a control, or rather, the original Web page. Version B is the modified version that contains changes intended to increase buying behaviors among visitors. Either through the use of user test scenarios or by introducing version B to the live website in a limited capacity, marketers can track increases in buyer behaviors from purchase sizes to increased page views.
A/B testing is meant largely as an evidence-based practice through which marketers can demonstrate the behavioral effect of website modifications on visitors. Consumers may respond differently to a wide range of visual and textual stimuli, which facilitates the need for research. From simply rephrasing segments of a website to drastically altering website navigation, changes can drastically affect buyer behaviors, and testing the changes can help website developers understand how users perceive the changes.
Knowing what to test
Knowing what to test for this practice of study depends largely on the objectives of the company and its website. For a retail website, you may want to increase customer response rates to email marketing campaigns. To do this, you would send one email to part of their audience saying to use a specific code, but the expiration date is phrased vaguely. Then you would send a different email to the remainder of your audience with a different code and a more specific explanation of the expiration date.
After sending these emails, you could then track the response rates for each email to determine which phrasing had the most direct impact on a recipient’s decision to buy. Companies like Amazon, Netflix and eBay constantly utilize this method in their marketing efforts to better understand the way in which users interpret deals and just on opportunities.
What can be tested for the purposes of this marketing research practice is quite numerous. Companies can test anything from simple Web page changes to phrasing of calls to action to new campaigns.
Knowing how to test
The key to effective A/B testing is reaching a significant portion of a company’s audience. In the event of testing out a change to a Web page, traffic should be randomly assigned to both the A and B pages. Each variant of the page needs to be exposed equally so as to effectively track buyer behaviors. Revealing the new changes to a smaller portion of the audience than the original change may not result in enough statistical proof to indicate any sort of change in buyer behaviors.
A/B testing tools
While most users can certainly test the effects of their changes internally through simple analytics, a variety of testing tools exist with a core emphasis on setting up A/B tests for a variety of simple to complex experiments.
Services like Google Website Optimizer, A/Bingo, Visual Website Optimizer and Unbounce all help in facilitating experiments for A/B testing. Utilizing software like these help focus on different areas of testing to understand what attributes of a website cause friction (reduce conversion rates), which discounts attract users and what areas of a website are most often navigated.
Most tools are offered on a paid subscription model and may require some integration with the existing website. Due to the obscure nature of A/B testing, few high-quality options exist to pick between for the purposes of optimizing a website. However, what applications that do exist can help in a variety of different capacities.