Split testing is conducting experiments to improve a site's metric and are also known as A/B testing or multivariate testing. Metrics are anything to do with improving a visitor's engagement to your website such as purchases, clicks, or form completions. Split testing is the equivalent to performing a controlled experiment: let's say you wanted to compare different variations of your website to see which will generate more traffic. Incoming traffic will be distributed between the original and your variations until a statistically significant difference emerges. Results will then be compared to determine with version has the greatest improvement.
Experimental Framework of Split Testing
Determine what needs to be improved.
Hypothesize the change.
Identify the variables that need to be changed and create variations.
Best practices of split testing include:
Focus on the Call to Action: Depending on your audience, your content might resonate more with one group or the other.
Aim for Global Maximum: What is the overarching goal of your website?
Elimination: Create fewer distractions.
Consistency: Make testing changes consistent throughout visitor flow.
Split Testing practices within web pages:
Layout: Size and arrangement of menus, forms, buttons, etc.
Visitor flow: How does the visitor get from A to B?
Text: Includes descriptions, headlines, content, etc.