Website Optimizer - A/B and multvariant testing

A/B testing and especially multivariant testing used to be available only to the big boys on the web that had the resources to use it but Google's free Website Optimizer tool levels the playing field and means every website has the tools available to find out what their visitors like best and what provides the best conversion rates.

Google tends to promote Website Optimizer as an Adwords tool but it can be used entirely independently of Adwords.  It is simply a tool that allows you measure the behaviour of visitors using different version of a page.

 

What is A/B and multivariant testing

In both A/B and multivariant testing different versions of the test page are shown to visitors and then a metric is measured that shows which is the most effective version.  Often the metric is the number of visitors who click through from the test page but can be something else such as the time on the page.  For the purposes of illustration, we shall assume here that our test page is a product page selling widgets and we want to see which variation results in the most visitors clicking on the buy now button and thus arriving at the first page of the buying funnel.

 

A/B testing

This is where two different versions of a page are shown to visitors.  They might be completely different pages if we are for example trying out a new look and feel to the product page and want to see if it performs better than the existing one.

 

Multivariant testing

This is much more sophisticated process.  Consider our product page again,  it is made up of lots of different elements.  Let’s just consider three: the headline, the body copy, the buy now button.  Multivariant testing presents all the different variations of each to visitors and then measures the results of which combination provided the best results. Even with just three sections being tested, the number of permutations rises very quickly, especially if there are a number of different versions of each element.

 

Using Website Optimizer

Data is gathered using javaScript tagging.  For simple A/B testing, adding the various tagging scripts is slightly more complicated than the usual Google Analytics tracking tags but not much.  Once the tags have been uploaded (and Optimizer checks that they have been done correctly), it is simply a matter of waiting for the results to come in. How long it takes to get a statistically meaningful winner depends of course on how much traffic the test page gets and much difference there is in behaviour of the various test variations.

But for a free tool, Website Optimize really is powerful. It was initially offered as a tool to improve Adwords landing page conversion rates – and incidently improve ROI and hence the willingness of Adwords advertisers to spend more – and if you are using Adwords then I would certainly recommend using WO to continually test and improve the effectiveness of landing pages.  But it has uses way beyond Adwords.  However your site generates its traffic and however it wants that traffic to navigate towards the site goal, then testing and measuring the most effective design, wording and navigation structure must be worthwhile. 

 

After all, the success of a site is a compbination of the level of traffic that it attracts and the effectiveness with which it converts that traffic and Website Optimizer is heaven sent to improve conversion rates.

17th October 2008

Search Engine Marketing and Web AnalyticsTraining Services