A/B Testing With Google Analytics Experiments and How to Make it Better
The most widely used and comprehensive analytics tool on the web, Google Analytics provides companies with data to help them make more-informed decisions. Through tracking metrics relating to your site and supporting A/B testing, it helps those who use it narrow-down the customer experience to what visitor’s desire most. Chances are that you’ve already got Google Analytics installed on your clients’ websites, meaning that the heavy lifting is already out of the way. The next goal is to get your first experiment up and running and understanding the best practices for getting powerful results.
Let’s take a brief look at just what exactly Google Analytics and GA Experiments are and then move on to how to implement a test and run it properly. One major drawback to running a test in Google Analytics is that it takes someone who has a bit of understanding on how to use it properly to get it right. After this section, you’ll be well on your way to running your first experiment.
First, a quick rundown on Analytics and Experiments
If you are already familiar with Google Analytics, then feel free to skip down a section (although reading about Experiments is recommended).
You’re probably well aware that Google Analytics is the most widely used analytics tool on the web. Its different metrics allow you to peer deeply into the two most important aspects of analyzing web traffic: where visitors are coming from and what they do once they find you. By now, you are probably very familiar with the drill–you create a Google Analytics account if you don’t have one, register the website you want to test, and paste the code.
Once you’re up and running, the good folks at Google start tracking a bunch of really helpful information for you, such as:
- Page per sessions
- Unique visitors
- Bounce rate
- Where the visitors are coming from
- Which language they’ve set their browsers to
- Any custom goals you want to track
- And just about anything else about them you could think of
The great thing about it is that this tool doesn’t just give you the raw data. It also gives you the opportunity to experiment with different aspects of your site to see how it affects that data.
Google Analytics Experiments
As a marketing agency or webmaster, you’ve got certain goals that you want visitors to achieve once they’ve reached your website. It could be purchasing an item, playing a video, or giving an email address. It’s all the same idea. A more detailed picture of visitor behaviour will help you narrow down your web content to exactly what helps convert the most visitors into customers.
With it, you can quickly set up A/B tests that find out which versions of a given page result in the biggest improvement of your chosen metric. In the simplest setup, you create multiple versions of the same page, and Google redirects visitors that land on the original page to one of the different pages. The traffic is initially split equally between the original and the variations, and then GA measures how they interact with your site having been exposed to a new experience. GA Experiments adjusts the proportion of traffic depending on performance as the test moves along (poor performing pages are phased out as the software becomes more certain about a variation).
Getting your experiment off the ground
Now that we’ve got the meta side of things out of the way, it’s time to start breaking a sweat. Let’s get the blood flowing by setting up your experiment.
Note: You need to have Google Analytics deployed on site (via GA code or GTM). More info on that can be found on Google’s support site.