A/B testing is the not so secret approach to optimizing any website. A/B tests compare two or more variations of a landing page by adjusting features such as text size, calls-to-action (CTAs), images, etc. It’s an ideal way to acquire data based on real consumers and make educated changes to your site.
A great tool for A/B testing is Optimizely. This intuitive platform allows users to make direct changes to any website without the need for developers. Optimizely has built in metrics and results tracking but also integrates with Google Analytics and other such tools.
Optimizely is a simple platform to navigate but some marketers may question which aspects of their site are important to test. See below for four of the most popular A/B tests to run on a website.
Checkout funnel distractions can cause customers to bounce before they complete a purchase. These include unnecessary CTAs, navigation panels, etc. Develop a test where the navigation header and footer are removed from the funnel to seamlessly guide customers through the purchase.
Homepage Banner Image:
Large banner images are a visual way to engage with anyone arriving to your website. The goal of this element is to drive clicks and optimize conversions. Run tests that compare model shots, product lines, single products, etc. to better understand consumer interests.
Bold brand colors are important to include when creating a CTA. Develop an A/B experiment testing different colored CTAs to determine which drives the most interaction. If your company is unsure about their brand colors, use Material Design to build a pallet.
Customize Pages Based on Campaigns:
A/B tests can be built to work alongside advertising campaigns on Facebook, Instagram, etc. Optimizely can send anyone who interacts with an ad to a variation of the website that contains the same messaging, creative, CTA, etc. Test this variation against the original page to determine if including ad features on the site results in increased engagement.
As a rule of thumb, Optimizely recommends running any test until there is at least a 90% statistical significance between variations before declaring a winner.