Split-Test.jpg

How To Run Split Tests In Google Analytics With Content Experiments

Leslie Samuel Analytics, Blogging Leave a Comment

Split Test

Man, I just read the title of this post and it sounded so technical. You may be thinking the same thing, but don’t worry – I’ll keep it simple with some step by step pictures. It’ll still be pretty powerful though. So lets go . . .

Split testing (or A/B testing) is one of those terms that you hear thrown around by marketers that’s very simple in concept:

  • You create 2 versions of a page
  • Half (approximately) of the people who come to that page will see version 1 and the other half will see version 2
  • Your split testing software will automatically calculate which page converts best (i.e. gets more signups, click throughs, or sales) and let you know.

Pretty simple – right? I know. However, as simple as it sounds, it’s pretty powerful. Lets do some simple math:

Lets say your Squeeze Page has a conversion rate of 40%. This means that for every 100 people that come to your site, you will get an average of 40 new email subscribers. Not too bad.

However, what if you do a split test and come up with a different version of your squeeze page that converts at 50%? That’s 10 ADDITIONAL email subscribers for every 100 visitors and 100 for every 1000 visitors. See – it adds up. AND, you can continue testing different versions to reach that 10% and increase from there.

Ok, lets take it one step further. Lets say you have a sales page that converts at 2%. This means that for every 100 visitors, you will make 2 sales. Lets say that like me, you have a product for sale that costs approximately $250. This means that for every 100 visitors, your 2 sales would equal $500.

Now, this is where it gets beautiful. You can double your income in 2 ways:

  • You can double your traffic
  • You can double your conversion rate

Now, I don’t know about you, but to me, doubling your traffic sounds like A TON OF WORK. No, it doesn’t just SOUND like a ton of work. IT IS a ton of work. Shoot, that’s a 100% increase in traffic.

However, doubling your conversion rate – not so bad. If you think about it all you need is a 2% increase in conversion. 2% increase in conversion is MUCH MORE doable than a 100% increase in traffic.

Makes sense right?

Ok, so it’s clear that split testing is a good thing. There are different services out there that allow you to run split tests. However, I’m not going to talk about all the various options, because they cost money. For this post, I’ll stick with the FREE option – Content Experiments in Google Analytics. Why? Because it’s free, very powerful, and does everything you need to do for simple split tests.

How To Setup Content Experiments In Google Analytics

For this walkthrough, I’m going to show you how we set up a split test for the Opt In for the 10 Free Videos Opt In page, where we tested 2 different versions of that page. The first version was a simple Video Opt In page with a video to the left with an opt in box to the left. Here’s a Screenshot of that version:

10FreeVideos

This was the page that I’ve been using since I created a squeeze page for the 10 free videos. I finally decided to start working on increasing conversions and got an even simpler design for an opt in page with a picture of a beach with an opt in form in the middle as shown below:

10 Free Videos

This one appealed to me even more but I didn’t think it would win because it didn’t have as much content as what was outlined in the video. But, I wanted to start with two very different designs just to see what would happen.

So, this is how this was done, and how you can do the same:

  • First, create the pages that you want to test. For me, that was the above 2 pages.
  • Login to www.google.com/analytics.
  • Select the account where you want to setup the experiment
  • In the sidebar, click on Content > Experiments
    Google Content Experiments
  • Click “Create Experiment”Google Create Experiments
  • Enter the URL of the page that you want to improve and click “Start Experimenting”Google Start Experiments
  • Step 1: Choose experiment pagesGoogle Choose Experiment Pages
  • Enter the name of the experiment that you want to setup and the URL of the pages that you want to test. You can also label the pages that you are testing so you can identify them easily.
  • Click “Next” to continue.
  • Step 2: Set Experiment Options
    Google Set Experiment Options
  • Select a goal from the drop-down menu or create a new goal. A goal is usually a page that the visitor is redirected to after they visit the pages that you are testing.
  • Select the percentage of people who will see the test pages that you created.
  • You can choose to receive notifications about important changes in the experiment.
  • Click “Save & Next” to continue
  • Step 4: Add and check experiment codeGoogle Add and Check Experiment Code
  • Use the Google Content Experiments Plugin for WordPress to insert the code easily to your pages. Once you install the plugin you will have a section under your page editor where you can simply enter that code as shown below:Content Experiments Plugin For WordPress
  • After you have updated your pages with the experiment code, click “Next” to validate the code, review your configuration, and run the experiment.

That’s pretty much it. Google will do the rest of the work for you and you can simply wait for the experiment to finish.

The Results Of The Test

I know, you’re probably curious as to which page performed better. So was I. Well, here are the results:

Google Content Experiment Results

As you can see, the winner of my experiment was the 10 Free Videos Variation, which was the one with the picture of the beach. The original version converter at 35.77% and the beach converted at 42.74%. It must be because beaches are so attractive :)

This is a 7% increase in conversion. Not only that, but now I can go ahead and test different pictures, headlines, and even other elements of that page to refine it even more and increase conversions further.

It’s a beautiful thing.

Your Comments

Have you been doing any split tests? If not, when do you plan on starting and what do you plan on testing? Let me know in the comments below.