Should I use Multivariate testing?

Short answer is yes!

Google provides the facility to test the conversion rates of more than one version of a homepage, however I needed something more specific for a particular client.

Basically the site was getting a large amount of relevant traffic through organic search and google adwords but the coversion rate could be higher. The site has a predominantly red design and we had the apply button in the same colour which meant that it did not stand out as much as it should. However it was very prominent so we didn’t think it the colour would have much impact.

We were very wrong.

What we did was created a simple cookie system generating a random number for individual visits to the site.  The system generats a number between 1 and 6 and assigns it to a visitor.  The system then selects a variant of the button based on this number.  We chose 6 very different colours that contrasted with the overall red colour of the site and also left the existing button in place as 1 of the 6 random numbers.

After a few hours there were around 100 instances of each random number recorded and there were no conversions on the existing red button.  However, there were 10 conversions on a yellow version of the button.  This is a truly staggering increase and shows the value of trying out various options on a site in a controlled way.

I look forward to seeing a larger sample set over the coming weeks so that we can extrapolate further findings and further increase the conversion rate.

Leave a Reply

Your email address will not be published. Required fields are marked *