Oops! We could not locate your form.
A guide to using Google’s Optimize. Covering what it is, why to use it, the tests you can run, and how to set them up.
Edit: Google Optimize will no longer be available from September 30 2023. There is some talk that it may be rolled into GA4 at some point, but this hasn’t happened yet.
Having a new website built, even with a proper UX design process in place, isn’t the end of the story for conversion rate. As marketers, we all want to squeeze out every bit of potential we can. As even a single percentage point improvement in conversion rate, can have a huge impact on revenue.
There is a whole industry built around this – conversion rate optimisation (CRO). There are numerous professionals out there who specialise in this, along with countless tools to help.
One such tool that is a great starting point for those dipping their toes into CRO, is Google’s Optimize. It also happens to be free, unless you desperately want the paid 360 version.
So in this post, let’s take a look at Optimize. We’ll cover why you would use it, the sorts of tests it can run, and how to use it.
Contents:
Optimize is Google’s very own platform for running optimisation tests, along with a couple of other features. It offers a nice simple interface and provides the tools for marketers to create, implement, run, and report on various types of split tests.
The platform can be as simple or as complex as you need it to be. With the ability to:
And a great deal more, depending on your objectives.
Split testing is a huge part of the CRO process. Split tests come in numerous forms, but at heart, they are about testing 2 or more versions of something against each other to see which performs better. The winner can then be implemented, and another test can be run.
As for the differences you can test, the choices are limitless, but should be data led rather than done on a whim. It is also better to test small iterative changes, gradually. So you can identify the specific contributors to improvements, or performance drops.
Just a few of the things you can test include:
Perhaps you have a landing page that isn’t converting as well as you expected, or maybe your CTAs aren’t pulling their weight. You might suspect that your homepage intro copy is not capturing your audience’s attention and contributing to a high bounce rate, or perhaps the current choice of imagery is a potential issue.
These are all things that can be tested. By testing, you can identify changes that will make a positive difference, without just making the change to your site, and hoping for the best.
Using an ongoing testing strategy, you can gradually make improvements site wide, that are based on data rather than supposition. These small changes can add up to significant improvements to conversion rate, and ultimately to revenue.
If you use Google products like Google Analytics or Google Tag Manager, then using Google Optimize is simple.
First you’ll need to create an account, which creates the container for your website. If you already have a Google account associated with your business, use this one. So it’ll be easier to connect with your linked accounts.
If you’ll be wanting to use the Optimize visual editor, which we’ll look at later, you’ll want to work in Chrome. There is a Chrome browser extension you’ll need to add, which will enable to use this visual editor. The platform will walk you through the process, or you can find more help on set up here.
You’ll then need to add the Optimize snippet to your website, or at the very least, the pages you wish to test. Depending on your set-up, you can do this through Tag Manager, or ask your developer to add it for you.
You can then connect Optimize to your Google Analytics account for measurement, and Analytics or Ad accounts, to automatically have your goals and conversions from those platforms as objective options.
Once this container has been created, you’ll be able to start building your tests (called Experiences). Your main container dashboard will contain all of your active, draft and expired experiences once you have some. Then each experience will have it’s own dashboard for set up, and reporting.
Optimize offers a range of ‘experiences’. These include:
You’ll be able to choose your experience type from your container dashboard. There is a blue button in the top right hand corner labeled ‘Create experience’. It will bring up this window for you to name your test, set the specific page you plan on testing, and select your test type.
Tip: Remember to use a meaningful name, so you can easily differentiate between tests if running several at a time.
So now let’s take a closer looks at the different types of ‘experience’ you can run, and how to set them up. If you work in marketing, the chances are you’ve heard of these test types before, but we’ll look at them specifically in relation to Optimize.
An A/B test allows you to test two or more versions or variants of a page against each other. Your original page, and then a variant or variants which are usually the same other than the particular change you wish to test. This can also be used to test entirely changed pages, but this doesn’t give you the insight into which specific changes drove improvements or declines.
Looking at all the data, you think your product page is underperforming, and you’d like to trial some changes that might improve the number of people adding the product to their basket.
Now you could do a full page change, but you wouldn’t learn much from the results about what made the difference. Instead you could run small tests, trialing changes to different elements, gradually improving the page through small changes evidenced by test results.
Initially you might test the colour of your add to basket button. When you’ve tested this effectively and have a clear winner or even see no change, you might then move on to button text. Then after that perhaps you could test your product imagery, or description, or the delivery information you provide on page. These tests will either rule out ideas, or confirm a winner, and allow you to make gradual improvements to your product pages, which will add up.
From the ‘Create experience’ window, name your experience, add the url of the page you’ll be testing, select A/B test, and hit the create button. You’ll be taken to the A/B test set up page.
Now you can get started by adding a variant (you can add multiple). Just click the blue ‘Add variant button’. It will create a variant and you can set a name. You’ll now see both the original and variant listed, and an edit button for the variant. This is where Chrome and that browser extension come into play. Hit the edit button, and you will launch the Optimize visual editor. It will show your original page, and offer a host of editing options.
Here you will make your change or changes, ready to test against your original. Click on the element you want to change, and then press the ‘Edit element’ button in the popup in the lower right corner. Then you can choose your edit type and make your changes.
In the screenshot below I have selected the H2 from the homepage, and selected to edit the text.
Then once you have finished making the change hit the ‘Finished’ button. You will need to save the variant with the button in the top righthand corner. Then you’ll be taken back to the test set up page.
Now you can work your way down the page, creating the conditions and objectives for your test.
First of all, you can add any additional variants and choose the percentage of traffic each variant should receive.
Next you can set up your targeting rules. You can keep it simple or get highly granular with your rule creation, depending on your objectives. This has two main sections.
Page targeting for creating URL based targeting rules:
Audience targeting for creating targeting rules based on audience demographics and behaviours, along with more advanced rules for things like JavaScript variables:
There is also a beta option for testing flows or funnels:
Next, you have a section where you can add a description of your test. Very helpful if you have multiple people accessing the Optimize account.
Then you can move on to the Measurement and Objectives section. If you haven’t already connected your Google Analytics account, you can do this here. Then you can choose your test objectives. You can select a primary objective, and up to two additional objectives.
Using the ‘Add experiment objective’ dropdown, you can select your objective type. You can choose from your existing Analytics goals, along with some standard preconfigured choices like page views, and bounces. Or you can create custom objectives.
Now you’re nearly there. Just the settings section. Here you can check you installation of the snippet it working correctly, set up any email notifications, select the percentage of your traffic you want allocated to your experiment, and set your activation event.
The latter is what triggers the test. You can choose page load, a custom event, or continuously. The latter can be useful for pages with dynamic content, but the visual editor will notify you of this as you are making your variant.
Now that you test is ready, you can scroll back up to the top of the page and launch when you are ready!
You will then be able to monitor your experiment, manage its running time, and end it from the Reporting tab at the top.
Phew! Ok, that was a long one, but now we’ve covered the first test and got the basics down, the others will take less time.
A Multivariate test allows you test page versions which have two or more sections changed, and all of the various combinations of these sections (up to a limit of 18 combinations). You can edit the names of each section and variant, to save things getting confusing.
Your landing page isn’t converting as well as you think it should. A multivariate test could be used to test a variety of landing page elements, and different possible combinations, to find which combination performs best.
You original page might have product copy at the top, a bulleted key benefits section in the middle, and a long contact form at the bottom. So you could choose these three elements as your sections. Then your variants could be different ways of presenting those elements.
Then your section variants could be:
Copy – a more conversational and shorter product copy section.
Key benefits – an image with each bullet point list header as one variant, and another with the key benefit lists placed in colourful boxes to help them standout.
Form – a shorter form with fewer fields as a variant, and another with the form nearer the top of the page.
Then the tool will test all of the possible combinations of these section variants:
As with A/B, from the main dashboard you’ll select ‘Create experience’, give it a name, the url to test, and this time select the Multivariate option.
The Multivariate test set up page looks mostly the same as the A/B one, except that the Targeting and Variants section looks a little different.
Once again you add your variants, and can add addition sections. Then you’ll see the the triple dot dropdown become available for each variant section, where you can edit to create your test changes, along with making duplicates for ease. You can have multiple variants for each section.
Once again, you’ll use the visual editor to create your variant section changes. This works just as before (remember you’ll need to use it in Chrome with the extension). Make sure you save each variant section you change. Using meaningful names for your variants and sections here will make your life much easier when looking at your reporting.
You can then move over to the combinations tab, and view and preview the various combinations that will be part of the test.
When all of your variations and combinations are made, you can go through the same process as before. Create the targeting rules, your description, set up your measurement and objectives, and then get your settings in order.
Then when ready, you can start your Multivariate test, and monitor and manage from the reporting tab.
A Redirect test is basically an A/B test, but perfect if you have created alternative page versions on your own website, and want to test these versions against the original. The variant will have a different URL, and the test will redirect a certain portion of your traffic to the original page (as decided by you), to the variant.
Based on research or discussions, you want to trial and new hero image on a key page on your website to reduce bounce rate. Rather than just going ahead and hoping for the best, you create another version of the page with the alternative hero image. You want to test these two against each other, so that you’re making changes based on actual data, rather than guess work.
Once more, starting from the main container dashboard, you’ll select ‘Create experience’. Once giving your experience a name, and page to test, you’ll choose the Redirect test option.
As you’re using your own alternative url, this test set up has fewer steps.
You will first need to make sure your variant page has been created, and is live. For best practice set it as ‘noindex’ so it doesn’t get indexed by Google, make sure you don’t link to the variant anywhere, and keep it out of your website’s menus.
Then you can get straight adding your variant.
You can add your variant in the standard way:
Or you can use more advanced options, using RegEx, or text replacement, along with parameters:
Then you’re back on familiar ground, adding your targeting rules, description, measurement and objective choices, and choosing your test settings.
Then, once again, you can start your test, and then manage and vein progress under the reporting tab.
Personalisation allows you to serve up more personalised version of a page to a visitor, based on certain targeting rules and parameters.
You may have a landing page for users coming from your paid social media campaigns. To make it more personalised and hopefully more likely to convert, you might want users who click though from Facebook ads to receive a slightly different version of the page to those from Instagram ads.
Using personalisation you could serve up a slightly different, more personalised page to each of those audiences. Maybe referencing the platform they came from, or perhaps showing a slightly different offer or call to action. Maybe previous research has told you that your audience on Facebook responds better to more conversational copy, whilst your LinkedIn audience prefers bulleted lists. This is also something you could do with personalisation.
Note: If you don’t have the paid 360 version giving you access to your analytics audiences, don’t worry. In the above example, you could create your targeting rules based on UTM parameters you can add to the destination urls of your social ads.
We’ve already gone through this bit three times, so you know the deal. Create a personalisation experience from the main container dashboard for your chosen page.
Use the visual editor to create your personalisation and save it. Then you can move on to setting up the targeting rules. Particularly in this case, your audience targeting. You’ll be given all the audience targeting options you have with the tests we’ve already looked at.
If we now go back to our personalisation example, we talked about using utm parameters to target those from an Instagram ad campaign, so that they’d receive the personalised version of a landing page.
To achieve this, you’d first create utm parameters for the destination link of your Instagram ad. This is good practice anyway for tracking campaigns.
You just need to make sure you include a unique campaign name, as well as the usual source and medium parameters. The unique campaign name is the important part for setting up this audience targeting in Optimize, as campaign name is the only parameter option to target by. So make sure it’s specific to the platform and the campaign.
If you use the same campaign name for your Facebook and Instagram ads, despite them having different source parameters, your targeting rule won’t work.
Aim for something like this, and make sure it is added to your ads destination url:
?utm_source=instagram&utm_medium=banner&utm_campaign=spring_sale_instagram
Then when creating your targeting rule, it’ll look like this:
Now getting back on track, you’ll go through the rest of the set up as usual, other than that you won’t need to create objectives. Instead the measurement section will give you an option on whether measurement in Analytics, and reporting are enabled.
Then you can launch your personalisation experience, and manage it as you do with the tests.
You want to add a banner to highlight a sale you are having on a certain range of products.
You’ll start this one off, as you did all the others by creating your experience. Except instead of being taken to a set up dashboard, you’ll be taken straight to the visual editor, where you can fully customise your banner.
When you’re happy with it, you can save and you’ll be taken to the familiar set up dashboard. From here you can then go ahead and set up your page and audience targeting.
The you can create your description, and as with the Personalisation set up, you can choose whether to enable measurement in Analytics, and reporting.
Then once you’ve completed the settings section, you can start your banner experience.
Now we’ve covered all 5 experience types, we’ll cover some of the important things to remember when running tests. These are less applicable to the Personalisation and Banner experience types.
Any experience you have running, in drafts, or have completed, will be accessible from your main container dashboard.
As with any experiment, sample size is key for reliable results.
The amount of traffic the page you are testing receives, will dictate the length and success of a test. The longer you can give it, and the more data you can collect, the better the result will be.
Basing a decision on forty visits and three conversions over a month isn’t really going to provide you with a meaningful outcome.
As mentioned previously, small, iterative changes are better than changing multiple things at once. Testing and refining each element will give you far more insight into what changes are actually beneficial, and those that aren’t.
You’ll end up with a far better optimised page overall, and come away with learnings that will help you optimise other pages on your site.
A ‘failed’ test doesn’t really exist. Even if there is no clear winner, or your original performs better, you have still gained valuable information.
Even a tiny improvement to your conversion rate can make a huge difference to your bottom line. You don’t need to aim for the sky. Be realistic.
If you’re looking to invest in a more comprehensive conversion rate optimisation strategy and want to find a CRO partner to help, get in touch. Our experienced UX and CRO specialists can dive into your data, plan and run tests, and implement positive changes.
Or if you’d like to better understand the user experience issues on your site, that could be effecting your conversion rate, take a look at our UX Review Website Audits. Our UX specialists will thoroughly review your website, and provide you with actionable recommendations to make improvements.
Discover the power of Enhanced Conversions in Google Ads. Learn how this feature can help you optimise campaigns while respecting user privacy.
For the majority of businesses, having a strong online presence is crucial.
Come July 1st 2024, access to Google Universal Analytics properties and the API will cease. With it will go any historical analytics data contained within your properties.
