We all want a website that performs, giving us more opt-ins and more sales whether it be through selling our own products or through promoting affiliate offers. What I’m going to talk to you about today is multivariate testing; a testing method that blows all others out of the water.
What is multivariate testing?
It is more commonly referred to as multivariate data analysis; simply put, this is just a testing method that takes into account multiple variables.
You’ve more than likely heard of A/B split testing which usually just compares two different sets of variables, but multivariate testing takes it to an entirely new level allowing you to improve conversions tremendously.
How to apply multivariate testing to your own website
I’m going to build on these posts by showing you 4 different tools you can use to start using multivariate testing to improve your own conversions.
The important thing to note is that this is a very basic testing tool in comparison to others out there, but it’s very straight forward to use and also free.
To get the most out of this you will need to use as many variations as possible, right now you’re only able to use the original page and test it against 9 variations.
I’d recommend creating a spreadsheet mapping out all of the variables that you’ll be changing across each version and then use several test groups where you can vary even more variables and continue testing.
This is another free tool; although it takes a lot longer to install, you do have greater control and don’t have to rely on Google Analytics.
You can either choose to use the free public service which involves adding some code or you can have complete control by installing it direct to your own server (although, this involves setting up a database and delving around with some code).
Despite the need for a code, if you want Genetify on your own server, the projects homepage has some very detailed instructions.
If you don’t want to use Google’s content experiment service but still want an affordable method to test your site then this would be a brilliant option.
This is a huge step up from GA’s content experiments, but I’d expect that since this is a paid tool; although, the multivariate testing only seems to be available on the more expensive packages.
Optimizely makes it easy to create variations and make any changes live on your site; you can utilize a custom goal tracking system and measure exactly what you need to measure. You can use this to conduct both A/B split tests and also multivariate tests.
If you’re new to blogging then I wouldn’t recommend starting here, but if your website is earning you a reasonable amount of money, then I would recommend you seriously consider this as an affordable and comprehensive testing tool.
This is an incredibly powerful tool, the price is a step up from Optimizely but this has some great features. Similar to Optimizely, you get the option for both A/B split tests and also multivariate tests.
There is a drag and drop interface for creating multiple variations so you don’t need to know any coding at all.
You also have behavioral targeting and geographical targeting options here, so you can show visitors different pages depending on their location, source and whether they use PC or Mac.
You may have seen some research in the past on clickmaps or headmaps (there’s a similar tool in Google Analytics). These are incredibly useful for finding exactly where your visitors are clicking and Visual Website Optimizer has this function too, but it doesn’t require any additional coding.
As there is a bit of a price difference between this and Optimizely, I’d suggest those with larger sites look at Visual Website Optimizer.
However you go about it, the best way to improve conversions on your site is to use multivariate testing, and Google Analytics provides an easy way to get setup.
We are all fighting against time which is why a lot of site owners tend to not test for conversions, but if you could add even as little as 10-20% to your websites income, would you give it a go just to see how well it could work?
The other thing to remember here is that habits change, what worked 8 years ago, may not work today, if you’ve noticed a drop in subscribers or noticed a drop in income while maintaining the same amount of traffic, then I’d recommend you try something like this on your own site. You may just be surprised at the results.
Have you had any luck with multivariate testing in the past? I’d love to hear your thoughts in the comments!