September 8th, 2014 | 3 MIN READ

A/B Testing the Untestable in Checkout

Written by author_profile_images Linda Bustos

Linda is an ecommerce industry analyst and consultant specializing in conversion optimization and digital transformation.

92% of merchants participating in the e-tailing group’s 16th Annual Merchant Survey ranked A/B, multi-variate or other usability testing as most or somewhat important merchandising tactic for customer retention (second only to web analytics).

While testing the small stuff like button color, headline copy, can get reasonable lift, often the most impressive gains come from bigger changes – new approaches to layout, radical redesigns that fix multiple conversion speed bumps and, of more recent popularity, responsive design.

And it’s often recommended to start A/B testing your checkout process before tackling other areas of your site, as visitors that get as far as your checkout are closest to conversion.

The trouble is, many marketers face limitations in fully testing the checkout process.

Both IT requirements and technology can be roadblocks. Even if you enjoy the luxury of dedicated IT, the time it takes to make such changes is costly.

Your testing tool and ecommerce platforms may only support simple changes like button colors, headlines, images, and relatively simple layout adjustments. Test that typically yield bigger results require more effort.

Testing platform limitations

Some A/B testing tools support WYSIWYG test page editing for simple elements like buttons and text, and enable blocks of content to be reworked inside dynamic panels. However, they still lack the capabilities to fully test any variable of the customer experience, for example, a radical reworking of the checkout flow.

Creating test variations within a CMS is an alternative, but you’re still limited to the content elements stored and managed by the CMS platform. Ecommerce elements are typically either hard-coded into the final checkout experience, or entire pages are loaded into the CMS and delivered to the end touchpoint, with ecommerce delivered in large, uneditable blocks.

It is possible to run ecommerce-specific tests at the ecommerce platform level, separate from your testing tool and CMS, but this is IT-intensive, difficult, expensive and difficult to reconcile with content testing and analytics.

Testing the untestable

One of the applications of our Cortex API technology to A/B testing is the ability to load discrete elements (individual cart lines, prices, checkout buttons, subtotal blocks, etc) as a service directly into the CMS as individual components. This means marketers can use them like lego blocks within a CMS to feed through a testing platform without IT, modifying them, moving or even removing them from an experience.

Why would you want to do this?

1. To test truly radical redesigns in your checkout (or any other part of your customer experience
2. To test different layouts for different touchpoints, like mobile sites and apps, tablets an in-store technology, which may be optimized with fewer or more “pieces” of content/functionality
3. To test where in your flow certain commerce pieces are served
4. To test in-checkout merchandising and promotions


Our ability to support more flexible A/B testing without the need for IT is one reason why Elastic Path has been named a major player in the IDC MarketScape: Worldwide Digital Commerce 2014 vendor assessment report, receiving the maximum score maximum for both integration capabilities and future integration strategy.

IDC MarketScape Worldwide Digital Commerce Applications Vendor Assessment is available for download (free).

Share on


Thanks for signing up!

You'll receive a welcome email shortly.

By submitting this you agree with our privacy policy.