As many of you already know, Google Website Optimizer (GWO) is a nice add-on to Google Analytics (GA) to run A/B/.. and multivariate tests for your pages. And while A/B test is pretty simple thing to do, multivariate tests are more fun.
However, working with some sites, you can't always specify the conversion page very well, the goal of your tests. If it's not an ecommerce one, or lacks a typical login / sing-up pair, you're probably hitting an "information resource". But such sites still need some work to be done, to get a better feedback and user retention. I would say, testing them for bounce rate and improving it later would be the nice thing to do.
Unfortunately, it's not very straightforward with GWO. You're still supposed to specify a page that will serve as conversion trigger.
One solution is that the bounce rate can be calculated by looking backwards, by looking at the conversions to any page. It's easy to do if you've got only one or two outbound links on your page. However, in the real life, it's unlikely to happen. Pages have dozens of outbound links. Going for all of them and putting a conversion script is probably doable, but such a headache.
But likely this can be done with a very simple Javascript. Let try to test the sample landing page. First of all, let's put the control script at the top.
Define the testing blocks.
And put the tracking code right before the
body.
The landing page is set up. Now, we can start with the conversion page. You can use the same one as conversion page in your GWO experiment. However, you don't need to add any tracking code to it (yet).
What we should do know, is to add a simple Javascript that will re-write all outbound links and will add an onclick action to them. In this action, Javascript will do the call to GWO and register a conversion. This simple JS should be put right before the closing
body.
[code lang="html]
[/code]
That easy. You might get an error while validating the conversion script, but you can easily copy & paste the content of script into file, and do offline validation for it.
Basically, that's it. You can finish setting up this experiment and run it. Here's how you will see results.
Have fun!
0 Comments