How to Run an A/B Test in Google Analytics
Google Analytics is one of the best tools to have in your SEO and site optimisation arsenal. The main function of Google Analytics is to provide site owners with information on traffic and user behaviour, but there are additional tools and functionalities that can be used to achieve different goals.
One of the ways you can use Google Analytics and its top features is for doing A/B testing. So, how can you run an A/B test in Google Analytics? Let’s find out, shall we?
Start Your Testing
It is easy to create a new report for running an A/B test on Google Analytics. First, you need to create a new experiment on your Analytics account. You do this by going to Behaviour > Experiments and clicking on the Create Experiment button at the top of the screen.
This will show a new screen when you can define some basic parameters for the experiment. The name of the experiment, the objective you want to achieve and the percentage of traffic to use as data sources in the experiment can be configured accordingly. You can also set up multiple metrics to track.
Hit ‘Save Changes’ and you are all set. You’re now ready to start defining the experiment parameters further and have an effective A/B test.
Configure the Experiment
Configuring the experiment is just as easy. Google allows you to add several variants you want to test against each other. You can even add custom names for each variant; this allows easier analysis when the report is generated later.
There are several metrics you can use to track the effectiveness of each variation. The best combination for A/B testing is to use Bounce Rate as the primary metric and then add other dimensions to the report as you see fit.
If you’re testing an ecommerce layout, for example, you can choose ecommerce from the dropdown menu. You can also track AdSense performance and other predefined metrics. As for Bounce Rate, you can access it through the Behaviour > Secondary Dimensions + Google/Organic > Top Page Views > Bounce Rate.
Fine-Tuning for Best Results
While the basic configurations of Google Analytics’ Experiment tool is accurate enough in most cases, there are times when some fine-tuning is needed. You can customise almost every part of the experiment to produce the best, most accurate results.
The confidence threshold is a good example. It is a parameter designed to tell the system how confident you are with the experimental page. A high confidence threshold will say that the winning variation is indeed effective against the others, while lower threshold suggests other factors need to be considered for a more conclusive result.
Configuring the experiment on Google Analytics is also not the only thing you need to do to start A/B testing pages and variations. You need to generate the experiment code and add it to the pages you are testing. Once the script is added, Google Analytics will start dividing traffic and measuring the predetermined metrics for the pages. You should start seeing the results in no time.