Split testing or A/B testing just got a whole lot easier!
The agony that was Google’s Website Optimizer has been transformed into Google’s Experiments and integrated into Google Analytics.
Have a look at the video above to see how to split test a web page in less than five minutes.
All you need is three pages
In order to split test (or A/B test) a web page you need three pages:
- An Original page – the page you want to test. Typically a squeeze page or a sales page.
- A Variation page – a variation of the Original page.
- A Success page – the page you want the visitor to arrive on if one of the first two pages has been successful.
A squeeze page is a page where you want the visitor to enter in an email address. If you want to test the efficacy of a squeeze page, you should create an identical page and change one important element in it – that’s your Variation page. And the Success page is the page where the visitor lands after clicking the validation link in the sign-up email.
It’s exactly the same process for a sales page. Copy a sales page of a particular product changing one important element (like a headline). And the Success page is maybe the check out page or the download page of the product (if it’s digital).
Google Experiments is in Google Analytics
I’m guessing you have Google Analytics. And, if you don’t have it yet, get it.
Experiments can be found under Content in Analytics. Click “Create experiment”.
Enter the URL or web address of the page you want to experiment on. This should be the Original page we specified earlier.
Then, once you’ve clicked “START EXPERIMENTING”, you should see a form where you can name the experiment and specify your Variation page.
Once you’ve done this you’ll, hopefully, see two thumbnail screen-shots of your alternative pages. Click “Next Step”.
In Step Two, you’ll need to specify your Success page or your Experiment’s objective metric, in Googlespeak.
This is a Goal. You can set up Goals in another section of Google Analytics (click Admin > Profiles > Goals), but this doesn’t matter because Google walks you through choosing a Goal in this step any way. Once you’ve pasted in the web address of your Goal you need to choose it from the dropdown and, again, click “Next Step”.
<head> section of your Original page. Click the “I’ll add the experiment code myself” radio button.
If you have the Genesis theme framework on WordPress this is quite easy as they helpfully include a “Custom Tracking/Conversion Code” field in their Theme SEO Settings. Otherwise you’re going to have to install the Google Content Experiments plugin to paste the code on the appropriate page.
Once you’ve updated the page go back to the Google Experiments and click “Next Step”. Google will then be “Checking your pages for working code”. Don’t worry if you get a rather annoying red message saying “No experiment code found” on the Original page. As long as you’ve checked the page source and seen that the code it there, you can click “skip validation and continue”. Otherwise, if you’ve got two green ticks you can go to the Next Step.
Then all that remains is to click “Run Experiment”.
An example of a sign-up box experiment
I once tested the colors on a sign up box in a squeeze page. Here was the winner:
Here was the loser:
The bolder, more contrasty, borders worked better.
An example of a squeeze page layout experiment
And, I’ve also tested the design of a squeeze page to see whether taking away a sidebar would increase signups.
And, amazingly, no! The winner was the one on the right. The page with the distracting sidebar got more signups. This was so unexpected I repeated the experiment and got the same result twice!
More about split testing
Read this article from the ever-amazing Robert Dempsey on split testing with Google Experiments and WordPress.