Recently I concluded an AB split experiment on a high traffic homepage. This site reached well over 2 million unique views in two weeks. The purpose of the test was to increase users landing on a product page.

While conducting the experiment we used the analytics package to track conversions all the way down the funnel. This funnel was a typical e-commerce website: Shopping Cart Page -> login / create an account -> billing and shipping information -> review order and confirm -> order thanks. In total the sales funnel consisted of 5 steps which are typical of an e-commerce website.

As the experiment progressed we started to see one of the homepages start to break away, reach statistical significance, and have a higher unique action rate to product pages (Total number of unique views to the product pages divided by the total number of unique views of the homepage x 100). We had a conversion of around 18% to the product page with the control (current homepage) and the winning homepage (challenger) had a conversion of about 20%. After we ran this 2 weeks, collecting plenty of data (minimal we want to run is 7 days to capture each day of the week. We ran this two weeks to capture each day twice to ensure no abnormal traffic trends occurred) we drew a conclusion that the challenger had a 2% higher conversion rate to the product pages and was chosen as the winner.

The next question that came up was why isn’t the funnel affected? Once we got past the product page and looked at the shopping cart the unique action rate for each step of the funnel was nearly identical. Even though the challenger had a higher conversion rate to the product pages the total successful orders were identical along with conversions along each step of the funnel.

Why?

This is a good example of why you must test an entire sales funnel. Only testing the homepage and thinking that will have an impact through the entire sales funnel isn’t logical. Sure, if one version of the homepage has “FREE Products, click here” and the other doesn’t, your conversions through the funnel may change. But if the price points and product offerings are the same on each homepage being tested and the only changes are calls to action, layout, etc., it’s just not logical to think it will have a higher conversion rate through the entire funnel.

The impact the homepage has is getting a user into a product page. From that point, the product page will have an impact getting a user to add it to the cart. Once in the cart, the cart page will get them to the next step, so on and so fourth.

When optimizing an on-line store always ensure to optimize the homepage, landing pages, and everything downstream through the funnel. Not optimizing the funnel is a terrible mistake.

At the end of the day all traffic will reach that funnel and you will want a fully optimized funnel on your website!

Recommended website optimization company:
Optimost.com