/ Ecommerce

A/B Testing for Ecommerce Websites

A/B testing or split testing is a relatively young tactic. In 2012, only 44% of businesses were using it. It’s the process of comparing two versions of a web page against one another with the goal of determining which version performs better. During the test, 50% of site visitors are shown Version A, and 50% see Version B. Your landing page exists in two distinct formats; a ‘control’ page, which is the original landing page, and a ‘variation’ page, which will be nearly identical except for the one change that you want to test.

Testing in this way takes the guesswork out of company decisions, bringing behavioural data to the foreground instead of basing important choices on presumed knowledge about your customer base. The success of proposed changes can be measured while the impact of those that don’t work is limited.

The results of A/B tests are evaluated, as you might imagine, based on whether the alternate page achieves a higher conversion rate.

Retail giant Amazon continually runs A/B testing on every aspect of the customer interface. When the company launched Amazon Vehicles, it even ran A/B testing on the colour of the car shown, with Version A displaying a grey car and Version B, a car with very colourful bodywork.

Benefits of A/B testing

Split testing offers massive potential for improvement. There are so many different elements to an ecommerce site that each site presents a multitude of opportunities for testing. You can experiment with one layout element at a time, even gradually affecting a complete site overhaul, or change the checkout process to add transparency or smooth the buying journey.

Because for ecommerce sites conversion rates directly impact on revenue, the results of A/B testing can be directly measured in terms of bottom line. These figures are not about lead generation, sign-up or brochure download, they are about sales and only sales. Successful tests can reduce cart abandonment rates and increase basket value. It’s even possible to analyse user behaviour to find out what interests your visitors and what doesn’t.

Dutch ecommerce site ReplaceDirect reduced cart abandonment by 25% using A/B testing. Responding to research that showed high or unexpected shipping costs as the top reason for cart abandonment, ReplaceDirect tested versions of their site that told the customer early on that there would be no shipping costs. The company also tested the checkout page, adding an order summary complete with total costs and delivery date.

A/B testing and conversion rates

Every single page on an ecommerce website prompts a desired action to be taken by visitors. Conversions include:

Making a purchase: This is the primary and most important conversion for an ecommerce store
Subscribing to a service
Submitting a survey response: Useful for customer feedback and advocacy
Signing up for a newsletter

For each webpage, A/B tests can be used to increase the conversion rate of the desired action. While simply increasing traffic can get results, A/B testing categorically optimises the page to ensure them. Pretty much any and every aspect of your site can be A/B tested, and sometimes the smallest change can yield incredible results.


How do I launch an A/B test?

Do your research: Study competitors’ websites and examine internal data to find areas that could potentially be improved with testing.

Based on your findings, create a hypothesis to be tested. For example, “Our shipping costs are not transparent and this is causing a high instance of cart abandonments, therefore if shipping costs are prominently displayed early on in the customer journey, the buyer is more likely to complete the purchase.”

Think of A/B testing as a scientific experiment. You must determine the variable and set proper controls to gain an accurate conclusion. In the example above, the variable is the visibility of shipping costs. By comparing conversion rates from the new version of the page with conversion rates from the original, ‘control’ page, you will gain an accurate result. Once you have selected the variable that you will test, all the other elements of the website must be kept identical or results will be unclear.

Don’t run your A/B test indefinitely; set strict parameters determining the length of time and number of visitors needed for an accurate, reliable answer.

A/B best practices

In order to get the most out of A/B testing, tests should be managed according to several ‘best practice’ principles:

Instead of a permanent 301 redirect, use a temporary 302 URL, since each test page is only live for a limited time.

Don’t use one set of website content for human viewing and another for robots. This technique, which is called ‘cloaking’ is used to avoid negative SEO impact such as loss of Google-search ranking, but will render any tests pointless, because the necessary 50/50 feedback from real customers and prospects is missing.

Run each test for only as long as is necessary to gather enough data to see whether Version A or Version B leads to more conversions. As soon as you have the information you need, implement any changes and move forward to another test.

Rezcomm is experienced in BI and ecommerce support. If you would like to speak to us about how we can help your business, contact us today for a chat.

Victoria Wallace

Victoria Wallace

Director of Digital at Rezcomm, Victoria specialises in Digital marketing, CRM, digital design, UI, UX, email marketing, airports, innovation, technology, travel, parking, eCommerce.

Read More