Chances are, you’ve heard of Google Optimize by now. It’s Google’s solution for A/B testing and personalization. It launched in beta in 2016 and left optimizers around the world waiting in line to try it out.
Now that it’s out of beta, you can give it a try without the wait.
But what can you expect? How do you configure it properly? How do you run your first experiment?
First, why Google Optimize over other testing tools?
Marketers love tools and tools love marketers. What results from this romance is tool overload. You have a tool for keyword ranking, a tool for broken links, a tool for social media mention monitoring, a tool for social media analytics, a tool for… you get the idea.
Google Analytics has been trying to diminish tool overload and bring marketers out from their silos for years. It addresses all channels, all conversions. It’s a central heart instead of multiple arms.
As Sean McQuaide of LunaMetrics explains, Google Optimize’s native integration with Google Analytics is what sets it apart…
Sean believes the deep integration allows for…
- Easier setup.
- More advanced targeting.
- More advanced reporting.
- Applying learnings faster.
It’s difficult to disagree that having Google Optimize data in Google Analytics and Google Analytics data in Google Optimize is a big competitive advantage.
7 Ways that Predictive Analytics Is Transforming Ecommerce
Predictive analytics help you understand what your customers are going to buy before they do.
If you’re reading this, you’re probably already using a testing tool like Optimizely or VWO. So, why give Google Optimize a try?
- It’s a familiar UI.
- Your Google Optimize data will be available in Google Analytics and your Google Analytics data will be available in Google Optimize, allowing for: more advanced targeting, more advanced reporting, more advanced conversion tracking, etc.
- It’s free, so what’ve you got to lose?
Google Optimize vs. Google Optimize 360 (Free vs. Paid)
I know I just said it’s free, but of course, there’s a paid version: Google Optimize 360. If you’re a small to medium-sized business or just getting started with a testing program, the free version will work for you. If you’re a big enterprise or have a very sophisticated testing program, you’ll probably need the paid version.
Here’s the official breakdown of the differences between the two versions…
So, to summarize, the limitations of the free version are…
- No Google Analytics audience targeting.
- Limited multivariate testing (16 variations).
- Pre-selected experiment objectives. Google Optimize 360 allows you to go back and change the experiment objective to see how the experiment would’ve impacted other Google Analytics goals.
- Limited concurrent testing (5 tests at a time).
Setting Up Google Optimize
Now, to get started, head to the Google Optimize site and click that big, green “Sign Up For Free” button.
Now you’re ready to create your account and container.
1. Creating an Account and Container
Choose an account name, first. This can be your domain, company name, whatever you’d like…
Google recommends opting into improving Google products, benchmarking and in-depth analysis. I recommend it as well since it’s a new(ish) product and the more info you can gather about it and how best to use it, the better.
Next, you need to add a container to your account. Your container name might be “CXL Blog” or “CXL Institute”, for example…
And you’re done! Now you should be looking at your Experiments view…
You can click the gray information icon in the right-hand corner of the view if you don’t see your onboarding checklist right away. Then you’ll see this on the right-hand side…
Notice that “Manage accounts and users” is already complete. If you expand the checklist item, you’ll see this message…
An account and container have been created for you! Google Optimize uses accounts and containers to organize your experiments. An account is the top level of your organization hierarchy, and it usually represents a company. A container is located within an account and typically represents a website.
You’ll also see your Account ID and Container ID…
2. Linking Google Analytics
Now Google Optimize will be encouraging you to start an experiment, but I recommend linking Google Analytics first. So go ahead and expand the second checklist item, “Link to Google Analytics”…
Click the blue “Link Property” button and you’ll be prompted to select a Google Analytics property…
Once you select a property, you’ll also be asked to select the view you’d like to link. Then just click “Link” and you’re all set.
3. Installing the Google Optimize Snippet
Now you need to install the Google Optimize snippet on your site. This is step three on your onboarding checklist…
Click the blue “View Snippet” button. You’ll end up seeing something like this…
Note that Universal Analytics (analytics.js) is required to install Google Optimize.
Now, you have two options for getting this Google Analytics tracking code updated: manually updating each page or using Google Tag Manager.
Free Advanced Google Tag Manager Course
Watch free courses on GTM and grasp the true power of Google Tag Manager's tags, triggers, and variables.
With Google Tag Manager
I recommend using Google Tag Manager.
If you don’t already have it setup for your site, Chris Mercer taught a live course for beginners via CXL Institute. If you have Google Tag Manager setup, but aren’t sure how to use it effectively, Jacob Shafer taught a live course for intermediate folks. I’ve also written a beginner’s guide. This is a really valuable skill to have.
Sean explains why he thinks GTM is the way to go, too…
So, head over to GTM and create a new tag. You’ll notice that Google Optimize is right there as a tag type…
Now enter your Google Analytics Tracking ID and your Optimize Container ID…
If your tag settings here don’t match your tag settings for your Google Analytics Pageview tag, you will run into issues. So, be sure you’re using the same Google Analytics Tracking ID input and the same “More Settings” inputs.
Now you can choose your trigger options. I’m going to experiment on my entire site, so I’m going with “All Pages”, but you can choose whatever you’d like…
Save it, preview it, debug it. And you’re done!
Without Google Tag Manager
Without Google Tag Manager, you can simply follow the instructions given after clicking “View Snippet”. You’ll be adding a single line to the existing Google Analytics tracking code. Unfortunately, you’ll have to do this page by page, which is likely to be time-consuming.
Are you familiar with the flicker effect? The flicker effect is when the visitor is shown the control quickly before seeing the correct variant. Of course, this has a number of negative impacts on both user experience and the validity of your test results.
Google created the page-hiding snippet to prevent the flicker effect. Just insert it as high as possible in your <head>. So, that’s between <meta charset> and your Google snippets.
Setting Up an Experiment
On to the fun stuff! Now you’re going to create an experiment…
When you click the blue “Create Experiment” button, you’ll be asked to enter the name of the experiment, the URL of the page you’d like to test and the type of experiment you’d like to run…
Perhaps you’re familiar with all of these experiment types. If so, just skip ahead to the Configuration section. If not, here’s a little about each.
This is the most familiar experiment type. You compare two versions of the same page to see which one performs better… A vs. B, control vs. variant. Visually, it looks something like this…
If you want to brush up on your A/B testing know-how, I recommend reading this massive, incredibly useful a/b testing guide that Alex Birkett wrote.
Redirect tests are a type of A/B test, technically speaking. Instead of testing two versions of the same page, you’re testing two separate pages against each other. This is useful if you’re looking to test a complete redesign or even two different landing pages.
A multivariate test allows you to test multiple variants of multiple elements at the same time to see which combination produces the best results. So, here’s how that might look if you were testing two headlines and three hero images simultaneously…
Alex also has an article on when to do multivariate testing.
For the sake of simplicity, let’s continue forward with an A/B test. You’ll want to start by adding a variant to test against the control…
Of course, that’s as easy as clicking +NEW VARIANT.
You can also change the variant weights and preview the variants here.
Google Optimize Visual Editor
Google offers a WYSIWYG visual editor, which should feel very familiar and intuitive to anyone who has ever used one before. (And you probably have… I’m using one right now to write this blog post.)
Here’s what you’ll see once you’ve grabbed the extension…
As I said, the experience is fairly straightforward and familiar. Here’s what you really need to know…
- The app bar at the top. Here you can change the experiment name and status, show changes, switch between variants, etc.
- The palette. This floats along as you scroll and contains all of the editable elements of your current selection.
- Current selection. In the screenshot above, I’ve currently selected the two lines of text before the bullets.
If you’re confused about anything as you get started, Google has a what’s what guide you can use.
When you scroll below the variant section, you’ll end up in the configuration section. Here, you can choose between either managing your objectives or your targeting.
First, choose your objectives…
You will be able to choose from basic objectives like pageviews, session duration and bounces. But what makes Google Optimize awesome is that you can also choose from any of the Google Analytics goals in your linked account.
In the free version, you can choose one primary objective and two secondary objectives. Remember that you can’t retroactively change these objectives in the free version, so be sure to choose all of the relevant objectives upfront.
You’ll also notice room to add a test hypothesis.
Here, you can choose the percentage of visitors to target and the weighting of visitors to target…
So, in this case, I’m targeting 100% of my visitors and showing each of my two variations 50% of the time.
Now on to the when…
Instead of explaining all of these targeting options in detail, as Google does at each of the pages linked to below, here’s a high-level summary…
- URL Targeting – Specific URLs.
- Behavior Targeting – New vs. returning, specific referral sources.
- Geo Targeting – Specific country, state, city, etc.
- Technology Targeting – A specific device, browser, OS.
- First-Party Cookie – Users that have a cookie from your site.
- Query Parameter – Specific pages or sets of pages.
- Data Layer Variable – Key values stored in the data layer.
This is where, if you had Optimize 360, you could do audience targeting.
Reporting is another area where Google Optimize really shines. Krista explains how that native integration with Google Analytics comes into play again…
(Both of Krista’s quotes in this article were taken from an article on her blog, which you should read if you’re looking to go beyond the basics after this beginner’s guide.)
But, if you’re keeping it simple and sticking to the Google Optimize reporting UI, here’s what you’re working with…
- Summary Card: Here you’ll see the experiment status and a summary of the results (so far). The leader, improvement, probability to be best, etc.
- Improvement Overview Card: Here you’ll see how your control compares to the variants based on the objectives you set. Note that you can click the column headers to have the results sorted.
- Objective Detail Card: Here you’ll see the performance of each of your variants against whichever objective you’ve selected from the drop-down list. Note that at the beginning of your experiment, the graph will show more uncertainty, but that uncertainty will narrow over time as more data is collected.
Here’s an example of the summary card…
And also the graph from the objective detail card…
Google Optimize is relatively new and it’s going up against giants like Optimizely and VWO, but the value of the native integration is hard to ignore. Especially with a $0 price tag.
At the very least, create an account and run an experiment. Hopefully this guide makes that process even easier for you. Then, see for yourself how it compares with your current A/B testing and personalization tool.
Is anyone using Google Optimize or Google Optimize 360? Please let me know what you honestly think of it in the comments.