How to improve your website performance with A/B testing

If your website is important to your business, you need to ensure that it delivers. Most people think they know what “looks good” and what “works” when they look at a web page. You’d be surprised how often your gut feeling is plain wrong!

Online marketing professionals know that the only way to really know what works best is to objectively test different versions of a web page. That way you’ll know for sure which version leads to more enquiries or sales. It is surprising what difference a small change can make, such as increasing the font size, changing the wording or the colour of a button. This is where A/B testing comes into play.

What is A/B testing?

A/B testing allows you to compare 2 (or more) versions of a web page to see which one performs better. Each version is automatically shown to a proportion of your website visitors, and you can then track which version led to more conversions.

The good news is – A/B testing is fairly easy to set up on most websites, particularly content websites such as most WordPress and Joomla websites, and you don’t need any expensive tools but can simply use Google Analytics to set up the experiment. 

Test example with button variations

In this example, we are trying to establish which type of button leads to more contact requests. The variations include size, colour and text. You probably would not test all of these variations at once but start with one variable, e.g. colour, then move on to size and finally text.


Prerequisites for running an A/B test

Here’s what you need to run an A/B test:

  • Enough traffic on your website to make the experiment worthwhile, otherwise it will take forever to get a conclusive result (consider Google AdWords if you need a temporary traffic boost)
  • Ability to set up variations of your web pages – this is fairly easy with most content management systems such as Joomla and WordPress, however it can be a lot trickier for eCommerce websites
  • Clear, trackable goal or conversion path – you need to know which result you want to compare, e.g. number of form submissions, online purchases, or even duration of page visits.
  • Google Analytics installation on your website so that you can actually set up tests and track your results

How to set up an A/B test

Without going into too much detail, the general process is to

  1. Set up one or more variations of your web page, e.g. by creating additional (hidden) pages in your content management system.
  2. If you are planning to track conversions such as form submissions or online purchases, make sure you set up goals in Google Analytics under Conversions -> Goals. Add the goal conversion tracking code to your confirmation pages, that way you will know how many times the goal has been completed from each page variation.
  3. Navigate to Behaviour -> Experiments in Google Analytics and follow the instructions to set up your content experiments.
  4. Add the Content Experiments tracking code to the pages you are planning to test
  5. Let your experiment run and review the results once you have reached a conclusive number of test results

Google Analytics Content Experiments interface:


Tips for success

  • Be clear about what you are testing, e.g. purchases or enquiries, and ensure that you can actually track these in Google Analytics
  • Don’t make too many changes to your web page at once, otherwise you won’t know whether it is the font size, image or button colour that made the difference
  • Be patient – unless your website is very busy, testing can take a few weeks or even months as you need a certain amount of traffic for the test to be conclusive
  • It is often best to just start with one variation of each page you are testing – the more variations you have, the longer it will take to get conclusive results

If you are interested in improving the performance of your website but not confident to set up the tests yourself, please give us a call – we can advise whether testing is feasible on your website and help you set up Google Analytics and A/B testing.