By Brianna Sullivan
At GNGF, we use data-driven website design to create brand recognition, deliver information and create business for our clients. While it’s easy to design it, launch it, and forget it, you should never be one to settle. At GNGF we analyze site performance data and look for ways to maximize a site’s usability and conversion.
Even with all the data in the world sometimes a subtle change in your layout, color, button placements, or badges can have an impact on your conversion rate.
But how do you know if a new design feature is going to work? How can you measure what site elements specifically influence site performance, and most importantly, conversion? This is where split testing comes in.
Split testing, also called A/B testing, is a controlled experiment developed to help determine a specific element’s contribution to a metric or goal. An example of a split test would be adding a button to the header of your site so that a random 50% of users would see the button, and the other 50% would not. Thus you are splitting your traffic into two buckets.
Literally once a user sees version A (with the button) they will be unable to find version B (without the button). Transversely if a user visits your site and is directed to version B they will be unable to see version A.
The beauty of split testing is that it allows you to hone in on the impact of a change to know exactly how adjustments influence user behavior.
Be careful. Let’s say you make multiple changes to your site: you add an additional homepage button, changed the position of your form, and moved your Avvo badge further down the page. If there is a significant change in your website data, how will you know which design adjustments took effect?
Split testing allows you to analyze significant changes to your site so you want to keep the changes to one or two at a time. This way you will be more confident of the exact impact on your website’s metrics before full implementation.
This type of experimentation supports purposeful changes in design strategy, and allows you to take more creative risks without affecting all of your traffic. Things to watch in your data include click through rate, bounce rate, time on page, and conversions.
If you determine a change has a significant positive impact then make it your new normal. Sometimes split tests have too little of a recognizable change in the data. If you test two alternate pages against 800 visits (or 400 visitors to page A and 400 to page B) and each pages shows similar data there is no need to make the change permanent.
When is split testing the right next step?
If you’re looking to grow and improve your web presence in a creative and dynamic way, split testing is an incredibly effective tool to add to your marketing strategy, but it isn’t the next step for everyone.
Here at GNGF, we recommend split testing when there is an item that we believe could be changed in order to increase conversions or on-site SEO. There is no magic number of visits, impressions, or other metric that tells us whether or not it is a good next step; however, we make sure that the website in questions receives a consistent amount of traffic. We don’t want to split test on a site with major fluctuations in traffic to preserve the integrity of the data.