A Step-By-Step Guide Creating a Split Test in Google Optimize:
Google Optimize is Google’s excellent free A/B testing and personalisation solution which allows you to create a split test in a matter of minutes. It is making online experimentation accessible to even the smallest website. Google Optimize is on the verge of doing what Google Analytics previously did for web analytics. Split tests are ideal for landing page optimisation and for responding to changing user needs.
This is important because Google Optimize gives webmasters and digital markers the ability to optimise their website design and experiences based upon evidence rather than subjective opinions. This could mean the end of the Highest Paid Person’s Opinion (HIPPO) as the primary driver of website design.
What is a split test?
A split or redirect test as Google Optimize categorizes them, is one of the most popular type of experiment used in digital marketing. As the name suggests, a split test is a type of online controlled experiment where incoming visitors are randomly allocated between two or more page URLs.
They are often referred to as redirect tests because they redirect a randomly selected number of visitors landing on the default page to one or more alternative URLs. This means you can set up a split test in Google Optimize in a matter of minutes. There is no need to make any code changes or use a visual editor to make changes to the HTML on your page.
Why use a split test?
A split test is a simple, but very flexible type of experiment. Some of the benefits of using a split test include;
- It allows you to test new page designs or content changes without the risk of damaging your conversion rate. By randomly splitting traffic between the default experience and the variant(s) you can have confidence that any statistically significant difference in performance is down to the experience and not other factors like traffic source or the time of day.
- It enables you to quantify the precise value of a new experience by measuring its impact on important calculated metrics such as revenue, sales or average order value.
- It helps identify when changes harm your conversion rate before you roll them out and allows you to learn from such mistakes without any serious impact on the bottom line.
- It minimises risk by ensuring changes are only rolled out when they meet your performance threshold.
- It allows you to respond to changing customer preferences by making it quick and easy to regularly test new features or designs on your website.
- It gives you the opportunity to be radical by running innovation tests where you completely redesign a page which suffers from poor conversion.
- It enables sites with high levels of traffic to test multiple variants against the existing page.
- It allows you to run experiments on a small proportion of visitors to the page (e.g. 10%) which can be helpful when you are testing a change on a vital and high-risk page (e.g. checkout or cashier).
How should you use a split test?
Split tests are ideal for measuring the impact on conversion between different page designs, but also you can use them for:
- A conversion focused alternative to your home page.
- Different hero images.
- Evaluate product usage intent by creating different headline, CTA and body copy combinations.
- Design schemes (e.g. visual elements such images, colours and rich media).
- Adding social proof (e.g. customer testimonials).
- Value proposition (e.g. your USP, headline and CTA).
- Location test (e.g. CTA and form placement).
- Registration page design.
- Product page design.
- Google Ads Landing pages.
How to create a split test in Google Optimize?
Provided you have your URLs for your different experiences you can set up a split test in Google Optimize in a matter of few minutes.
1. Create a Google Optimize Account and Container:
If you haven’t already set up an account go to my post How to set up and run experiments in Google Optimize to find out how simple it is to create a Google Optimize account. This explains how you create an account and set up a new container for your website experiments.
2. Create a new experience:
Go to the container for the website where you intend to run the split test. Click on ‘Create experience’ on the far right of the screen and it will take you to a screen where you can define the nature of your experiment.
Here you will be asked for:
- Name: Use a suitable prefix to distinguish between an experiment (e.g. T for test) and a personalisation campaign. For example, T005 is our fifth A/B test.
- What is the URL of the page you would like to use? Enter your default URL here.
- What type of experience would you like to create? Select ‘Redirect test’ as this is what Optimize calls a split test
- Click: ‘Create’
3. Add Variant:
This will take you to the ‘Draft’ experience where you can ‘Add variant’ and edit the default page URL if required.
By clicking on ‘Add variant’ you can give your new experience a name and paste in the URL address of the new page. Use the ‘Advanced redirect’ mode to modify fragments of a URL by swapping, changing or adding strings to a single or multiple URLs.
You can now add further variants if you wish. But be careful to remember that each time you add another variant you will also be reducing the proportion (weight) of traffic going to each experience. Unless you have lots of traffic you may be best to limit yourself to a single challenger variant.
4. Page Targeting:
Here you can use URL targeting rules to test multiple variants of a landing page. For example, use a sample product pages as the Editor URL (the default) and then create targeting rules to target all ‘/product’ pages to evaluate a change in the template design.
5. Audience Targeting:
Carefully selecting the audience of your experiment can be crucial to its success. You don’t want returning users to react in shock to your new experience because they remember seeing the default design. To minimise this effect, it’s wise to target new users and consider other rules to ensure the experiment is relevant to your audience.
For our test shown at the top of the post, it might be appropriate to use Geography to target cities nearby as our variant experience is promoting the consultancy as a ‘local’ business. Your audience targeting should reflect the nature of the experience and what your hypothesis is trying to prove. But, also be careful not to be too specific as otherwise you may not have enough traffic to complete a test.
Now you can give a description to your Google Optimize redirect test. Make sure this is self-explanatory for anyone who views it in future.
7. Link to Google Analytics:
If you haven’t already linked your experiment to Google Analytics, go back to your container view, and select ‘Container settings’. Here you can link your experiment results to Google Analytics to ensure that you can analyse results data in GA.
Set your objectives by either using a pre-populated list of metrics or creating a custom objective.
Provided you have linked to your Google Analytics reporting view the list option will have many of the most common metrics you might need. This will include GA goals that you have set up, ecommerce metrics, AdSense Revenue, bounce rate, pageviews and many other standard metrics. As I had previously set up a goal for ‘Contact Us’ form submission I used that as my primary objective.
Custom objectives allow you much greater flexibility as you can create an objective from an individual event (e.g. lead generation form submission) rather than being restricted to GA goals and standard metrics. You can also use the number of pageviews for a specific URL as an objective.
Set a number of secondary objectives if necessary so that you have a good understanding of how the new experience changes key metrics. This can help identify if there are conflicts between different metrics (e.g. increase in returns vs rise in average order value).
The settings section allows you to validate your Google Optimize installation is working correctly before you launch your test. Further, ensure you enable email notifications so that all stakeholders are automatically informed when a change to an experiment is made.
If your experiment is high risk or stakeholders have concerns about the impact of the test, you can adjust the traffic allocation here. Personally, I believe this is counterproductive as it just means you have to run the test for a longer period before you can identify if the experience is having a negative impact on your conversion rate.
The activation event allows you to define what triggers your experiment. In most cases this is a page load, but you can be more sophisticated and use a custom event. However, a custom event requires a piece of code to be added to your website to inform Google Optimize when the event occurs.
10. Star Experiment:
You should now be ready to begin your Google Optimize split test because all the stages in creating your experiment should now be displaying a green tick.
11. Edit Your Live Test:
If you launch your Google Optimise split test and then realise that you need to amend it in some way, don’t worry, as you can now edit an experiment without having to stop it. This is a great new feature of Google Optimize that many of my fellow CRO consultants have been crying out for a number of years.
12. Reporting Split Test Results:
One of the great things about Google Optimize is that it uses AI to recommend what you should do about your experiment. The Reporting tab in Google Optimize provides you with a wealth of information about the performance of each variant and gives your clear advice on when you should end the test and which, if any, variant is the clear winner.
Google Optimize offers you the ability to create sophisticated split tests in a matter of minutes. Split tests are a fantastic way of helping you to optimise landing pages, sign-up forms and many other high volume pages. Respond to changing user preferences and improve your conversion rate by using split tests to measure the performance of different page designs. It’s that simple, you will wonder why you haven’t done it before.
- About the author: Neal (@northresearch) provides web analytics and CRO consultancy services and has worked in many sectors including financial services, online gaming and retail. He has helped brands such Hastings Direct, Manchester Airport Group Online and Assurant Solutions Ltd to improve their digital marketing measurement and performance.
- Neal has had articles published on website optimisation on CXL and Usabilla.com. As an ex-market research and insight manager he also had posts published on the GreenBook Blog research website. If you wish to contact us please send an email to email@example.com. You can follow us on Twitter @conversionupl, see Neal’s LinkedIn profile or connect on Facebook.