What is Split Testing?


In case the terminology is not on your radar, I’ll begin by revealing that the wonderous, the truly splendiferous, process of split testing is carried out by phizz-whizzing, whoopsy-splunker web designers who know what’s what when it comes to boosting conversion rates (there’s no prizes for guessing what I saw at the cinema last night!). In more detail, (and I’ll try and be sensible), the process is used to compare different versions of the same webpage in order to identify the page that performs best. With the results to hand, web designers can determine whether changes to the website will have a positive impact on conversion rates or not. (I’ll explain conversion rate optimisation in a moment).

Split Testing/ A/B Testing/ Multivariate Testing

Before going any further, it’s important to note that split testing may also be referred to as A/B Testing, or multivariate testing. While both terms fall under the split testing umbrella, these processes have key differences; A/B testing is the most basic split test which compares version A (original) against a version B (variation), whereas multivariate testing enables you to test multiple changes at the same time. It really depends on your goal as to what method is best to use. For example, if you are considering major, back-end design changes, split testing is the best method to use, whereas for simple, front-end changes (like changing the colour of a button) A/B testing is preferred.

Conversion Rate Optimisation

When we talk about conversion rates we are referring to the percentage of visitors who end up reaching a given goal, i.e. when a visitor to your website takes an action that you want them to take such as making a purchase, signing up for an email newsletter, downloading an app, or any other type of online activity (‘call to action’ in marketing terms).

A conversion rate optimisation plan is carried out during the early stages of the split testing process to identify pages with low conversion rates, or high-drop off rates that can be improved upon. Google Analytics, Kissmetrics or other analytics services can help with analytics reports and insights via click through rates and on-site metrics as well. With this data you can create an informed test hypothesis which may cover changes such as the colour of a button, navigation elements, adding a button or link or something entirely custom.

Split Testing Software

Deciding on the most effective type of split testing software can seem daunting due to the amount of options available to choose from, however, to give you an insight, here are three of the most popular tools available:

Google Content Experiments: this free service is offered via Google Analytics and lets you split test variants within the programme.

Visual Website Optimizer (VWO): this is similar to Google Content Experiments, but in addition includes a what-you-see-is-what-you-get (WYSIWYG) editor. This means that you can make changes to the page for testing purposes without having to change the underlying HTML or CSS code. VWO offers a free 30-day trial.

Optimizely: this is similar to VWO in that it provides a WYSIWIG editor so you can make changes easily without having to alter the HTML or CSS code.

Testing/ Results

When the test is run, the traffic is randomly split among the different versions to see which version performs best. By tracking the performance of the webpages, web designers can implement positive design changes based upon the accurate, informative data. This takes the guesswork out of website optimisation and should put you in good stead for creating a whoopsy whiffling website (sorry, couldn’t help throwing one last BFG reference in there) with all the attributes needed for achieving a healthy return on investment (ROI). And we all want a whoopsy wiffling website, right?

Phizz-whizzing – brilliant
Whoopsy-splunkers – fantastic
Whoopsy wiffling – Great
(Roald Dahl’s BFG)