Definition: A/B testing, or split testing, presents 50% of users with an alternate version of a webpage in order to test the effectiveness of a particular variable. A/B tests are evaluated based on whether the alternate page achieves a higher conversion rate. To ensure that results are not impacted by uneven sample sizes, it’s important to evenly split both page versions among all visitors (hence the term “split testing”).

AB Tests and Conversion Rates

Every page on an ecommerce website promotes a desired action to be taken by visitors, and A/B tests aim to increase the conversion rate of that action. Conversions include:

• Buying a product – the primary and most important conversion for an online store
• Subscribing to a service
• Signing up for a newsletter
• Submitting a survey response

A/B tests aim to get the most out of existing traffic by optimizing a web page to promote conversion. Increasing traffic can garner results, but A/B testing aims to optimize a page to increase revenue with current spend.

On-page elements that online businesses commonly split test

Virtually every aspect of a website can be A/B tested, and even the smallest changes can sometimes yield surprisingly beneficial results. Some website elements that are commonly tested include:

• Calls to action: Button shape, color, and placement; CTA text.
• Images – Video clips, audio testimonials, and hero images
• Body content: Primary and supporting content — either above or below the fold — can impact conversion. Simply switching out one word may significantly increase conversion rates.
• Social proof: Testimonials, badges, partner logos, and reviews are commonly A/B tested in various combinations
• The navigation structure: Some pages perform better with fewer navigation options on page, while others increase in conversion when users have easy access to supplementary content.
• The check-out process: All of the above elements can be tested throughout the checkout, and many stores also tinker with flow itself. SaaS ecommerce platforms are often highly optimized for checkout, enabling marketers to concentrate their A/B testing efforts elsewhere.

A/B testing is also used to tests prices, promotional offers, optimal trial-offer lengths, the effects of free delivery offers on sales and more. Whenever online stores want to gauge the effectiveness of a feature of marketing technique, and A/B test can help them assess it.

How to Launch an A/B Test

1. Research: Look at competitors and examine internal data to look for areas that could potentially be improved with testing.

2. Hypothesis: Based on your findings, or even a gut feeling, make an assertion to be tested. For example, Users aren’t sure where to find ‘Shipping info’ on the homepage; therefore, if it were prominently displayed on the top navigation, the conversion rate would increase.

3. Determine the variable: Remember that an A/B test is a type of scientific experiment and must be done with proper controls to gain an accurate conclusion.

4. Set parameters: Use an A/B test calculator to determine the length of time and volume of visitors needed for a reliable test. All website elements besides the variable being tested must be kept identical to yield valid results.

5. Launch the test: Use Optimizely or another A/B testing platform to begin the test and record your findings.

Best Practices for an A/B Test

For best results, A/B tests need to be managed according to proven principles such as the following:

• Avoid cloaking — the use of one website content for human viewership and another for robots to view — to avoid negative SEO impacts such as demotion of a site’s Google search-engine ranking.
• Use a temporary (302) URL redirect (instead of a permanent 301), since it is only a limited-time experiment.
• Only run the test as long as needed to gather the data necessary to make a solid choice between version A and version B. Once enough data is compiled, time is best served by implementing any necessary changes and moving on to another test.

Learn more about how to increase revenue with split testing from the Bigcommerce Blog: 3 Proven A/B Testing Steps