A/B testing is one of the most important things you should be doing for your eCommerce site. In eCommerce, your site’s layout, design and feature set are a crucial part of the product – not just the physical products you sell. Yet many businesses don’t place high importance on testing. Without it, you’re essentially displaying products without ever knowing how your customers respond to your layouts.

A/B tests compare a variable against a control to determine which one is more popular. For eCommerce sites, elements of control page “A” are compared to the new version “B” to find which one increases interest in the page. This could be in the form of click-through rates, conversions, pages viewed per visit, or whatever goal is specified.

Say you want to encourage more people to click a button on your page to sign up for a service. To test the effectiveness of the button, you’d use the current page/ button design as your control and a test it against a new design. Both designs would be released simultaneously to a customer sample, and their clicks during the test period would be tallied to see which button drove more clicks. You’ll usually see a pretty clear indication of one particular design that users prefer. A/B testing is a simple, yet effective way to determine which version “wins”.

With the results of individual tests and the collective results over time, testers will get actionable data about what design, layout and features users prefer.

But repeat after me: “one design does not fit all”. User demographics and preference play heavily into what works for one site but not for another. Even though your site may sell something similar to another, your audience may be different and users may behave in very dissimilar, contradictory ways. It’s important to test your site specifically, and not necessarily follow others in your industry.

Instead of grasping at straws when deciding on design or feature changes on your eCommerce site pages, consider A/B testing to help give clearer, data-driven insights that can provide concrete recommendations to improve your site’s performance. The only thing you should not A/B test is the decision to A/B test.

Have you been shocked by results from A/B tests? We’d love to hear what you learned.