How to Set Up and Run AB Testing with Google Analytics

Use Google’s Experiments to enhance your website’s engagement and conversion metrics


Last week, we’ve shared a few tips for A/B Testing, but how should we start one?

We apply all the web design best practices but without any visible performance improvement. You’re not sure of what it is, and you need to know how to apply more continuous testing to maximise the engagement and conversion metrics of your website.

In this article, we’re going to explain how to set-up a split test inside Google Analytics in a few minutes.

The goal is always the same: make them get what they want as fast as possible while getting them closer to one of your conversion needs.

Google Analytics comes with a basic experiments feature that allows you to compare different variations of a page and split the traffic accordingly between them. Keep in mind though that you’ll need quite a bit of volume to get statistically significant results (rule of thumb is to have min. 100 conversions but this can vary quite a bit) and be able to decide if the changes actually worked or not.

If your traffic numbers are still low, the most common way to generate cheap traffic fast is through Facebook ads. In our next article, we’ll share with you a deep dive on how to set up and run them the most effective way.

But How Can You Run an A/B Test in Google Analytics?

Setting it up takes only a few minutes. Once you decided what you need to test, you can get started.

A.) Get Started

Under the BEHAVIOR tab, you can find the Experiments tab. If this is your first experiment, , you’ll probably see it like this:

Setting up A/B Test in Google Analytics
Click on Create Experiment on the top left.

B.) What should you experiment?

Name your experiment with whatever your objective is. Here is where you can set a detectable outcome to check results and see what’s the best variation possible.

AB Test - Create New Experiment
Here you might:

  • Select a Site Usage data (like bounce rate, shown here)
  • Select an existing goal (like purchases, opt-ins etc)
  • Create a new objective or goal

This depends on what you’re testing in the first place. You’ll be surprised by all the metrics related to your website. You can see the bounce rate, for example, on Behavior > All Pages. On our test we’ll use Bounce Rate as Objective.

Google Analytics behaviour reports

By default, all those advanced options are off and Google will always “adjust traffic dynamically based on variation performance”.

Let’s keep going?

C.) Configure your experiment

Now you add the URLs for all the pages you need to test the variations. Just copy and paste the links, like this:

The New Google Analytics Content Experiments

You can give names so you can remember them easily. I’ve named them like the above.

D.) Script code

Editing the page’s code can look scary at first but it actually quite easy. The first thing you see under this section is a helpful toggle button to email the code snippet to your developer.

Setting up your experiment code
AB Tests step by step in Google Analytic, adding script code to your page

If you’re doing it yourself, make sure to double check the pages you’re testing to make sure that your default Google Analytics tracking code is installed in all of them.

Next thing, you highlight and copy the code provided. You’re going to look for the opening head tag in the ORIGINAL variation (which will be located on the top of your HTML document).

Search for <head> to make it easier for you.

Once that’s done, click NEXT STEP in Google Analytics and they’ll verify if everything is ready to go. If it’s not, they’ll let you know.

Inspect element in Google Chrome
Experiment code validation

And… voilà!

Don’t forget that you can only make an important decision once your experiment reached minimum 95% statistical significance. You can measure it using this very helpful tool.

What we can learn from this is that websites are never 100% finished! We always need to experiment with new ideas and analyse data to keep increasing our conversion goals.


No Comments

Sorry, the comment form is closed at this time.