What is split testing?
Split testing, sometimes referred to as A/B testing, is an important aspect of optimising landing pages, websites and digital ads for maximising ROI. Which conversion objective is being tested for is completely dependant on the nature of the businesses goals; clicks, calls, form submissions or purchases are all things, which split testing aims to boost.
Split Testing involves creating measurable changes to the particular aspect that is being assessed and comparing the performance of such variations against a control. This process is frequently carried out by advertisers and marketers to empirically evaluate measurable elements of a website, page or advertising campaign.
How Does Spilt Testing Work?
The process of split testing involves running two or more versions of a landing page, website or ad; a control version (the original) and a version which has had one or more of its elements altered in order to find out which performs better. Incoming traffic is evenly and randomly divided between the control and the variant(s) without visitors knowing in order to produce objective results. Once enough traffic to produce a statistically significant end result has visited the landing page/website/ad, testers will be able to infer whether the variations have or have not improved the subjects’ performance based on metrics such as click-through rate and conversions.
The non-exhaustive list of things that split testing is used to test is;
- Visual; pictures, graphics, videos, colours of text and CTAs etc., the layout of different visual elements, size of buttons, text, menus and forms.
- Textual; headlines, call to actions, and any copy.
- Visitor experience; how users are able to navigate the page to get from one point to another.
Among many, one of its major values is that it can evidentially settle the difference between subjective web page or ad features such as colours and visual design.
How can marketers split test to generate meaningful results?
One of the most important things to have before you start your split test is a data centred objective for your (or your client’s) particular business objectives. What works for one will by no means always work for another and so it’s imperative to formulate a hypothesis based on your specific requirements prior to initiating a split test.
Within the hypothesis, you will need to determine the volume of visits to your control and variants that will produce a clear result about which one performed better. For example, you may decide that a 90/10 split is statistically significant enough for you to identify the page or ad that produced a higher conversion rate.
Following this, create the relevant modifications that you intend to test and ensure that all confounding variables, such as traffic sources, are accounted for. Likewise, ensure that all links, CTAs and visual elements are identical across all versions to avoid generating inaccurate results. Finally, direct traffic to the relevant pages or ads and wait for the results to come in.