In this blog post, I am going to explain how you can set up your first, simple, Google Analytics content experiment. Before you set up your first Google Analytics content experiment, you need to know the following.
- The goal of your experiment. E.g. To reduce the bounce rate of the page, or to increase opt ins, or to increase sales.
- The single action you want the visitor to take. E.g. Click on the ‘Buy Now’ button.
- Your hypothesis. The reason behind your test. E.g. The layout of the page is too cluttered. It is distracting to the viewer preventing them from taking the desired action. If the page was less cluttered, the visitor will be more likely to click on the buy now button.
- Details of the change you are testing. E.g. To declutter the view I will remove excess links, add more whitespace and use a single column layout.
Next, you need to create the page you want to test. If you have a WordPress site you can use a plugin to clone the page. When creating pages for testing you must set the canonical tag of the test page to point to the original page. Again, if you are using WordPress this is easy to do if you are using an SEO plugin.
Now you are ready to set up your experiment.
Create a new experiment
In your Google Analytics account, go to Behaviour -> Experiments and click on ‘Create Experiment’.
Set up your experiment – initial settings
- Give your experiment a name that relates to the objective, or goal of your experiment. For example, decrease bounce rate.
- Set the objective from the drop down menu. Your objectives should be related to hitting more of the goals you have set in your analytics or to engagement such as decreasing bounce rate, increasing page views or increasing session duration.
- Decide on the percentage of traffic you want to see the experiment. This will depend on a few things. How fast you want to declare a winner. If you want faster results then you need more traffic so set this to 100%. This means all your traffic will be shared across your original page and your test page. If you are testing something that you feel is very risky and you don’t want all of your traffic seeing it, then you can set that number here.
- Decide if you want to be emailed about how your test is progressing.
Since we are focusing on simple tests, you can leave the advanced options in their default state. Click on ‘Next Step’.
Configure your experiment
- Select your original page and give it a name.
- Check the box “Consolidate experiment for other content reports”. This keeps your analytics data clean by attributing all traffic to your experiment page to your original page.
- Select your test page and give it a name.
Again, we are only creating a simple experiment so don’t add more than one variant. Click on ‘Next Step’.
Set up your Google Analytics experiment code
Click on the button ‘Manually insert the code’ and you will see the code you must insert after the opening head tag of your original page. How you add code to your pages will depend on the platform you have built your website with. You can also send this code to a developer who can install the code if you are unsure. (Check back soon for an update on how to do this with Google Tag Manager)
Review your experiment and start it!
Checking your experiment
Now your experiment is running check back on it every few days.
Go to Behaviour -> Experiments and click on your experiment. You are looking for when a winner has been declared.
If no winner has yet been decided you will see a notice that says “Experiment running – no winner yet”.
The experiment will end when either the time limit has been reached or one page clearly outperformed the other. When the experiment has ended you will see a notice that says “Ended” and the reason for ending the experiment. That could be time limit reached, winner found or no winner.
Please come back to this page soon because I will be adding how to run experiments in Google Optimize. This is a tool that is in beta at the moment. I am running my own tests on it to see how it works.
If you need help setting up your experiments or help with defining what your tests should be please get in touch.