Split Testing Our Question and Answer Site was a Real Surprise

I’d never really thought about split testing until recently. It’s a relatively simple process whereby you offer 2 variations or more of something to your target audience and let THEM decide which one they like (Yep, imagine that. Letting your customers tell you what they want…)

Before, I used to use a lot of blue on my sites. You see, I like blue. And I used to use clever, corporate style headings and titles because I wanted my business to be perceived as bigger than it was. I liked that too. But my mistake was to assume that because I’d been an avid student of marketing, with more than a couple of accomplishments up my sleeve, that I knew what was best for my customers. But relatively low sales proved that I was very, very wrong in this assumption.

Enter split-testing. For our Question and Answer Site offering at Qhub.com, we decided to employ some very cool split testing software called Visual Website Optimizer. It gives us a snapshot of a live page on our Question and Answer Site, let’s us experiment by moving, removing, re-colouring and re-wording certain elements, then splits visitor traffic between those variations and measures what visitors to that page actually do.

In one of our most recent tests, I wanted to know what combination of elements on the Qhub.com home page encouraged visitors firstly to click on our Get Started button to create their own Question and Answer Site, and secondly to actually complete the registration process and see an up sell page to a discounted version of our software. The results were a real surprise.

Here’s a snap shot of the original Qhub home page (http://qhub.com/) before split testing where we tested about 2000 visitors, mixing combinations of the Take the Tour button, the brown voicemail tab at the edge of the screen, and a testimonials box at the bottom of the page which isn’t shown in this shot :

Qhub.com home page before split testing

Now here’s a shot of the test results, showing that while Combination 8 that showed visitors no voicemail tab, no Take the Tour button and no testimonials box resulted in a 71% improvement over the standard page for clicks on the Get Started Now button, it resulted in a nasty 60% decrease in the number of visitors who actually completed the registration process and saw the up sell page :

Qhub.com Comnibation 8

By comparison, Combination 7, which showed visitors no voicemail tab, no tour, but showed the testimonial box, resulted in a massive 131% improvement for clicks on the Get Started Now button, plus a 62% improvement in the number of visitors who actually completed the registration process and saw the up sell page:

Qhub.com Combination 7

Now, you’ve probably heard how effective displaying testimonials on your web site can be, but I think you’ll agree that a 131% improvement is really something. I had, in fact, expected that the losing combination, with less distractions for the visitor, would be the winner, but my intuition was incorrect, and our Question and Answer Site was under-optimized as a result.

So, if you have a web site and you’re interested in what your customers want, I highly recommend you use a split testing tool like Visual Website Optimizer and learn that intuition is no substitute for exact data on what your visitors want from you. I’m so sold on the idea that I now split test everything, even posts like this one!