Google Optimize is a testing tool that is included within the Google Analytics package, providing you with the opportunity to test various elements on your website. Most individuals and businesses will already have access to the free version of this, or alternatively there is the option to upgrade to the full 360 suite. Despite the hefty price tag that comes with Optimize 360, the additional features allow you to be able to conduct more tests at once and have more advanced targeting functionality too.
No matter which of the options you have, there is still the opportunity to make great improvements to your website. So, with that in mind, we have created a basic guide for those marketers out there that have Google Optimize available, but haven’t yet utilised it.
Once you have linked your website account, you will see that tests are split into numerous sections. Experiments are categorised into live, draft and ended – which allows you to easy be able to assess what is live at any one time on your website, and what is in progress. In this section you will be able to see a maximum of 5 experiments live at any one time with the free version of optimise, however with Optimize 360 you are able to run over 100 at any one time.
Before delving too far into the building of your experiment, it is important to first create a hypothesis of what you are wanting to achieve. Is there a business question that is never answered, or something that you have always wanted to change on your website to see if it performs better? Well now is your time to do it, in an experiment you can review the performance as it runs, making sure that you don’t implement anything that may have a negative impact on your website KPIs.
Once you have developed a strong hypothesis, you then need to start thinking about what your variations will look like, and what metrics you are going to measure the test performance by.
When it comes to creating your experience, it is important to:
- Assign a suitable and consistent naming convention
In our experience we ensure that we include the start date within this to make it easier to keep track of, and also assign each test a specific number.
- Consider the URL you want to make changes on and preview the experiment
Ensure that you have thought through the page(s) that you are going to be testing on, and how these can be grouped together.
- Be aware of the different types of experiments you can run
There are numerous options of tests that you can create within the tool, so you need to decide which is the best fit for the activity that you are wanting to do. Your options include:
- A/B testing
- Multivariate test (MVT)
- Page redirect test
Building the test
Once in the draft stage, there are capabilities to add as many variations as you want or require. But bear in mind that you want your results to reach confidence, so going crazy with too many variations may not be beneficial, as you will then need to run your test for longer. Always review your traffic volumes, to help you identify how many variations are possible.
The variations always have the weighting split evenly between them as standard in the tool, but if required you can edit this.
There are numerous ways in which you can target where your test goes live, the simplest being ‘URL starts with’ or ‘URL contains’. These capabilities allow you to be able to target a specific page, a group of pages or all pages on the site.
In addition, you are also able to apply audience targeting, which can range from basics such as device targeting, or looking at the type of visitor (i.e. new or returning visitors). Alternatively, you can do more advanced testing by using specific query parameters. Mostly, the basics of the targeting capabilities allow you to have a solid start at experimentation on your website.
One of the main advantages of Google Optimise is its ability to easily target PPC campaigns – which is really valuable as you can conduct landing page testing by linking specific PPC accounts and campaigns. This is something that we often work with our paid media team with, to ensure that their activity is effective, cost efficient and leading to on-site conversions.
Once you have linked up your Optimize account to the correct GA view, you are all set to start adding in objectives. These should be focused on what you are trying to achieve. Applying these objectives means that optimise can review the performance while the test is live, so you do not need to do this manually. You are limited to 3 objectives within the free version, but with some thought and forward planning – 3 objectives are efficient.
Using the edit button next to each of the variations allows you to be able to modify the code for each. There are a couple of ways that you are able to do this:
- WYSIWYG (what you see is what you get) – You can edit the core functionality and styling of the page without any development knowledge. This can be done by clicking on each of the features and editing the code. We would advise you being cautious if you take this route and ensure that you do additional QA checks to make sure that this has had the desired effect.
Before putting any tests live, always ensure that you QA the test to make sure that the targeting is correct and that the variations are displaying in the way that you expect. There is preview functionality available within optimise that is a good start, as it allows you to be able to preview on different sized devices, i.e. desktop, tablet and mobile.
For an additional layer of confidence, we also test with URL fragments or by adding query parameters, which allows us to do an additional layer of QA on different physical devices and browsers. We always ensure that we cover the top percentage of devices and browsers for each of our clients, to eliminate any potential risk.
Once you are happy and completed all the above steps, you can put your test live, monitor and watch the data start to collect!
Top tips for beginners testing in Google Optimize
- Once your test goes live, make sure that you check that nothing breaks. This can be done by going into the variations yourself, or also checking that the data in the reporting view looks accurate.
- Ensure you do not communicate the ‘winner’ with stakeholders too early. Your test should reach confidence and statistical significance so that you can be sure that the test can be attributed to the changes in performance. It is also important that you have a large enough sample size in the pot that has been tested. We like to use tools such as the CXL calculator to make sure that we can confidently say which variation was successful, and if it should be implemented on-site.
- Do not be put off testing if your ideas aren’t always successful. It is important to keep learning from the tests that you do, and eventually you will be able to learn more about what does and doesn’t work for your users.