What is Multivariate Testing or MVT?
Multivariate testing is a method of testing multiple combinations of content and call to actions on a website and measuring the results. It provides factual data on what people on a website prefer in terms of layout, call to actions and other content.

Why would I do this – I know what is best?
Do you really? Everyone thinks they are a web expert these days, and there are absolutely some people who live for usability on websites, but every site is different. Customers will react based on the branding, layout, whether they’ve just been paid, what they are looking for – the list goes on. MVT allows you to test different variations and use the one which people on your site use the most.

OK, so how do I get started?
Google Website Optimizer is one of the easiest ways to get started. The experiment wizard gives you instructions on how to install the javascript required, and you then create variations which Google will serve and track the success rate of the original, and the various combinations.

This is all great, but can you give me an example?
That I can. The biggest success I’ve seen to date, is changing a simple button to say “View” instead of “Book”. This increased the numbers of clicks on that button by just under 50% – something which I was amazed at – but the figures say it all!

Key points for MVT

  • Don’t start with too many variations, especially on low traffic pages
  • A negative result isn’t bad – it shows you’re currently using the better combination – but try another different one
  • MVT is constant – don’t stop with the first success story, try and improve on it further
  • Allows you to base decisions on fact instead of assumption, and remove problems like design by committee

I’ll touch on this again at a later date and come up with some examples of how I’ve been using it.