A/B Split Testing

The most basic form of randomised split testing, great if you're just starting out with Conversion Rate Optimisation (CRO).

Want to know which version of a web page is best? Convinced that your idea will increase the level of enquiries from your site? Use A/B split testing, which can tell you with a high degree of scientific certainty which page performs better.

How Split Testing Works

A/B split testing 'splits' the traffic you receive to more than one version of a web page. A proportion of your visitors are sent to the original (or control) page and the others are sent to a different version of the page (with Google Website Optimizer you get to determine what the proportion split is).

It's called A/B because your visitors either get page 'A' or 'B' and it's called randomised testing because the traffic split is random.

Over time, once you have received enough visitors to each page version, A/B split testing will tell you which page version results in more goal conversions. Google Website Optimizer presents the results through its easy-to-read dashboard. The goal of A/B split testing is to establish a winning version - once you have achieved this you can adopt the winner permanently.

You can use Google Website Optimizer to setup and manage A/B split tests, it is remarkably easy. Keeping tests simple to start with makes things much easier.

Commit to Continual Improvement!

Once you have adopted (or kept) the winning version get started on a new experiment to improve your goal conversion rates. If you stick with experimenting you will see conversion rate gains. It's worth bearing in mind that you can also run A/B/C split tests - where visitors are split between 3 different pages - but this type of test gets slightly more complex (and requires more web traffic)

Use A/B Split Testing if:

• You want to improve your site conversion rate
• You believe in statistics and scientific methodology as the best way of improving your site
• You want to get started with CRO testing
• Your site is getting less than 1000 visits a week

There is a great deal of science and maths that goes into establishing which page version is the best. We won't go into it all now but if you get in touch we will be more than happy to discuss what is meant by 'statistically significant' and how it will work for your site.