The Maxymiser blog has moved!

You should be automatically redirected in 6 seconds. If not, visit
http://www.maxymiser.com/news-and-events/blog/
and update your bookmarks.

Tuesday, March 20, 2007

Multy-whatty Testing?!?

Multivariate testing is a frightening sounding term but in reality, it’s quite simple. What it does is take all the guesswork and old wives’ tales out of creative decisions for websites. We’ve all been there; the web designer says the "add to basket" button should be on the top right, marketing reckon bottom right and the MD gets involved wants it on the left!

These arguments can (and should) crop up daily in any online business that’s going places. There’s no way of telling who is right without testing because although bottom right will work for some sites, top right might work for others. Try asking your web guy to set up a head to head test of all three positions and the resulting workload is likely to make him just give in and side with the MD’s suggestion. Even if you do go ahead and set up a three way test, there’s a big technical labour overhead with setting that up to display each location of “add to basket” button to display to 33% of visitors then you’d have to program a back end to report the stats. A month later you might know the outcome and be able to implement the best performing button across the whole site. In the meantime, you’ll have lost potential sales due to not having the best possible location on show.

Maxymiser takes the stress out of finding the content that gives the most conversions, registrations or other key metric. Your web coder simply needs to implement “Maxyboxes” on the page so that the Maxymiser server can insert the three different content variations being tested. If you have a high traffic site, it will only be a matter of days before cast iron results from our statistically-sound sample will be available ready for you to either implement yourself or to simply click a button to have Maxymiser implement the best performing button location.

That’s the simple view, in reality, most of our clients want to test many combinations of a large number of site elements. Next entry, I’ll take you through how Maxymiser adds value with Continuous Optimisation and how we can help with your design of experiment to produce statistically valid results in a shorter time with Taguchi Orthogonal Arrays.