What's an A/B Test?
A/B Tests are very common in Conversion Rate Optimization (CRO), and it's easy to understand why. They are used to test variations on an interface in order check how effective a specific change can be for a certain purpose.
A/B tests are called like that because there is a control version of the interface, and one or several variations. Users are randomly presented with one of the variants, and their behavior is tracked to see which version performs better.
Thanks to browser cookies, each user sees the same version every time they enter or refresh the page; either the control version or a variant.
Here's an example: Let's say you want to increase the conversion rate of one of your landing pages. You wonder if testimonials could help to do that, and where you should place them in the page, or the style those testimonials should have. This is easy to see running an A/B experiment. You can create a variant of the page that includes testimonials at the top of the landing page; then, another variant with testimonials just before the CTA button; and leave the control version without testimonials. This will allow you to see how many people actually click on the button and convert on each variant of the page to see what works better.
There are many popular tools to create A/B tests out there, such as VWO (Viual Website Optimizer) and Optimizely. Google used to have its own Website Optimizer, which was integrated into Google Analytics in 2012 and called Content Experiments. Still, it wasn't as powerful as other services.
Google Optimize is a brand new product. At first it was only available for Google 360 customers, but now they've decided to make it available for free. Although the free version is still in private beta for now, so you may have to wait and join the waiting list before being able to use it.
How to set up an A/B test on Google Optimize
Before starting: You need to do this from Google Chrome, because at some point you'll be required to install the Google Optimize Chrome plugin.
Firstly, go to optimize.google.com to enter the main Dashboard. Now it's empty, but in the future you'll find an overview of your "Running" tests.
There is a list of steps to go through on the right, called "Container Information". You have to complete those steps in order to start using Optimize.
The first step is to "Create account and container". You will have to create one container per website, so that you can run tests on it.
Right now, you can only build one container per account. It's soon to tell, but this restriction might continue in the future, at least in the free version of Google Optimize.
After naming the container, the second step is to link Optimize to the Google Analytics View that corresponds to the website where you'll be running the tests.
What's a View? Check Google Analytics' Hierarchy.
The third step is to implement the Optimize snippet. It has two different parts. The first one is a script tag that includes your current Google Analytics snippet, plus the Google Optimize plugin. It's just one more line of code inside your Google Analytics tracking code.
If you are already using Google Analytics on your website, look for that piece of code and add the missing line.
The Optimize line should look like this: "ga ('require', 'GTM-ContainerID')";.
There is also an optional but recommended snippet, a page hiding snippet. It hides your web page before a variant is applied, to minimize page flicker. If this sounds good, copy and paste that code as early in theas possible, certainly before your Google Analytics tracking code.
What is page flicker? When some elements of the page change after it's loaded, it can be because of a poorly implemented A/B test. In an ideal scenario, you don't want your users to see what the page looks like without the variation.
You can also add Optimize to your Google Analytics via Google Tag Manager. Go to "New Tag" > "Tag type > Google Optimize", add your Google Analytics Tracking ID and your Optimise Container ID (which was in the line of code you had to add to your Analytics tracking code).
On the fourth step you'll finally start creating your experiment (yay!). Add the URL of the page you'd like to test, which will be called your "editor page", and select the type of experiment you'll be running.
Types of experiments:
- A/B test. It presents each user with a different version of the same page to track their behaviour.
- Multivariate test. For more complex variants, with several independent sections.
- Redirect test. Add a second URL and Google Optimize will send the user to one address or the other.
For a first time testing the platform, we recommended you to go for the A/B Test type. If you select it, you'll go to a new page to work on the experiment's draft. The first thing to do there is to "Add Variant". As you'll see, all your variants will appear on a list. The original page is already there by default. You can add as many variants as you want.
Click on the little arrow icon in a variant's like to "Go to Editor" and add changes to that variant.
This is when you'll be asked to install the Google Optimize plugin in order to continue.
After you install the plugin, try to edit your variant again. It will take you to your page, where you'll be able to make editions using the plugin. This is arguably the best part of Google Optimize, because of how intuitive and powerful it is.
On the top bar of the editor you can also select different devices to preview your changes on them.
A couple of tips:
- Ask yourself what you want to learn from this experiment. Don't start making random changes. Each change should have a clear purpose and help you optimise your site.
- For really insightful tests, make one change at a time. If you make a variant which is completely different from the control version, how will you know the impact of each of the changes you made? If you want to test several things, make a plan and run several experiments over time.
- Give your variant a descriptive name that tells you what you exactly what you changed and what was the purpose of the test. For example: "Increased font size - Time on page".
When you're finished, SAVE your experiment. Next, you'll have to define which are its objectives, and describe them. This will enable Google Optimize to track and measure the impact of your variant. A Google Optimize objective is basically a Google Analytics' *Goal ** . If you haven't set any goals yet, you'll need to do that first.
Read more about setting up goals in Google Analytics.
That's it! Click on the "Play" button to start running your test and collecting data. On the overview of your variants you will be able to see the percentage of users who are presented with each variant. If you check more details, Google Optimize will let you know for how long an experiment has been running, how many users have seen each variant, and how well it is performing according to your goals / objectives.
You can also see this data on Google Analytics.
See you on Facebook and Twitter. Send us a message if you have any further questions or comments on the matter.