Split testing is also known as multivariate testing or A/B testing and is a means for conducting a controlled and randomized experiment to improve website metrics. These metrics include click-through rates, purchases, form completions or any other desired action. Split testing works by “splitting” incoming website traffic between the original page and a page with some variation designed to improve the metric being tested. After a statistically significant amount of visitors have arrived on both pages the metrics are compared to determine which variation has the best results.
Split testing is frequently used in testing the changes to registration pages, sign-up forms and calls to action, but can be used on any part of a website where some measurable action can benefit from improvement. One such case is the use of A/B testing during the checkout process to decrease abandoned orders.
While many website changes can seem subjective, the use of split testing can make these changes objective, since the data collected will allow for the measurement of actions, and can either support or refute a hypothesis regarding what will work best.
Split testing also allows variations to be tested that target specific groups of people to tailor a more personalized experience. For example, testing might be done to optimize conversions on a webpage for males or females, or for a certain age group or geographical region.
Split testing subjects are nearly unlimited as almost any element on a webpage can be the subject of an A/B test. Marketers may want to test some of the following elements:
- The layout, such as the arrangement or size of buttons, forms, links and menus.
- Visual elements, such as pictures and videos, header images, or even the colors of the site.
- The flow of visitors on the site, or how they navigate from page to page within the site.
- Any of the text on a site, such as headlines, descriptions, or calls to action.
When conducting a split test there are some best practices that should be kept in mind and adhered to, including:
- Calls to action should be a priority. Different text has differing meanings to different people. Some marketers test dozens of variations to get the best possible conversion.
- Always keep the goals of the entire website in mind when testing, not just the individual page. Sometimes it might be advantageous to send people from one page to another for better conversions.
- With the above in mind, always keep changes consistent throughout the site to improve conversions at every step of the visitor experience.
- Keep a minimalist mindset. Cluttered pages distract users. Clean pages with minimal elements typically convert better because users may only have one or two choices, rather than several.
Split testing is not a once and done activity. Rather it should become a habit, and performed consistently to improve user interaction and activity. Every visitor to a website is a potential customer or conversion, and split testing can help ensure that the greatest number of visitors become customers. Continual testing is also important because tastes change, and your visitor demographics can also change over time.
Split testing is the experimental part of your web design, using a controlled methodology that delivers better conversions, and a deeper understanding of your visitors and eventual customers. Split testing was developed from direct mail and print advertising campaigns, and has been found to be extremely useful for web designers and marketers.