Multivariate testing with GWO in WordPress
This article is about Google Website Optimizer which is no longer supported since 2012 and is only preserved as archive.
If you want to make the most of your blog in WordPress, you’ll have to find out what works best for your readers.
How to do this?
Well one thing is to ask them. An easy implementation of 4Q survey tool will help you get clear answers. The task completion rate coming from the three questions in this survey will definitely help you improve.
However, taking the advice of Jacob Nielsen “Don’t Listen to Users” to heart we need an alternative solution. This is where Google Website Optimizer comes in. Google Website Optimizer allows you to test what it is that you want to know. In two ways. A/B testing and multivariate testing.
For A/B testing you can get a clear explanation from John Doherty (no worries, the plugin works fine 🙂 ), so that leaves multivariate testing.
In addition to the A/B testing plugin, this requires one additional plugin. It´s called WordPress PHP Execution Plugin and it allows you to write php code in your WYSIWYG editor.
Why you need this?
The trick is…
To do this with a subtitle in your post would look like:
I´ve used this to track click outs of links within a post (here’s how that works) and it works like a charm! Enjoy!
4 Replies to “Multivariate testing with GWO in WordPress”
Hey man –
Love the post. Thanks for linking to mine, and thanks for the clear explanation about multivariate testing! I want to do more testing on my site soon and this will help.
Glad you like it. Why repeat good instructions when you can simply add to them 🙂 Curious to hear what your tests uncover!
Question: What alternative can you recommend now that Google removed MVT from their testing tool since they move GWO into Google Analytics?
Depending on your preferences there’s a long list of paid alternatives to GWO. When it comes to ease of implementation I’m familiar with Visual Website Optimizer and I’ve been introduced to Sitespect (but unfortunately didn’t have the chance to try it).
As an alternative to actual multivariate testing (if you’re looking for free), I’m seeing the option to create up to 9 variations with Experiments in Google Analytics. While this doesn’t show you which part of the variation had the most impact, it does provide conclusive data on which variation to use. An added benefit is the endless segmentation with regard to the other data in GA.
Hope this sheds some light on alternative options.