Digital marketers love tools. My career is built on search marketing, a specialty known for its fondness of tools. For years, we have had unique tools with unique metrics for each part of our job. One tool to check and report rankings. One tool to check and report broken links. One tool to check and report authority. Lots of data and all of it living in its own universe, separated from the rest of the marketing mix.
To add to the complexity of this system, each tool is often 'owned' by different departments. Marketing may control goals and reporting, creative may control experiment setup, while development is tasked with making things function properly. Sometimes the left hand has no idea what the right hand is doing.
That system of intricate data silos worked for many years. We stitched the data together the best we could and made decisions knowing that while one hand delivered a report the other held an asterisk.
Google Analytics is changing that. It acts as the glue that holds our data together, allowing marketers to get a firm grip on KPIs and hold each digital marketing channel accountable for the capital it's allocated. Google Analytics strives to position itself as the hub of all digital marketing activities. It's working to be more than just a place to collect data and report numbers, it wants to be the place where stories that drive internal and external communications begin.
Enter Google Optimize, an A/B testing and personalization tool that uses Google Analytics data to power your CRO efforts. Obviously A/B testing is nothing new, neither is serving personalized content based on customer behavior. The true progress here is how Google Optimize pairs with Google Analytics, and how easily we can tie our experiments to KPIs in Google Analytics.
For an introduction to Google Optimize, I wrote a handy blog post to shorten your learning curve. The rest of this discussion assumes that you are familiar with the basics of Google Optimize and why marketers use it. In this article I focus on how Google Optimize and its enterprise version, Optimize 360, will change your life by putting everything under one roof.
Native integrations between Google Optimize and Google Analytics offer benefits at nearly every turn. From setup and design to reporting and iteration, the familial ties make website testing and personalization more successful. And you don't have to remember any additional passwords to do it.
Why use GTM? Event tracking, that's why. Google Optimize uses GA goals as experiment objectives and pulls data from GA to calculate experiment results. So if you want to test objectives that involve user interaction, you'll need to set up an event-based goal first. The easiest way to do that is by using GTM.
Want to track if a landing page variation causes more people to download a PDF? You'll first need to track downloads as an event using Tag Manager (easy to do) and then set up an event based goal in GA. Once the goal is set up inside GA you'll be able to test against it on Google Optimize.
Often, you're already using goals to track if users are reaching important pages or completing important actions on your site. Other A/B testing tools require you to replicate these goals, to the best of your ability, inside their testing interface, opening up the possibility of two different definitions of the same ultimate goal.
By using Google Analytics goals as test objectives, you can be confident that the results of your experiment will impact your analytics goals the way Google Optimize says it will. Remember the goal: consolidate your data and make it work for you.
Who you test is as important as what you test. You might have the best test idea in the world, but if it isn't relevant to half of the people that see it then you're not going to see the results you expected. That's why targeting is so important.
If you've used other A/B testing platforms, then you can expect much of the same targeting options here. Experiments can be setup to target URLs (host, path, query parameters, etc), Behavior (referrer and time since first arrival), Geography (city, state, zip, etc.), and Technology (browser, OS, device, etc).
For the Google Tag Manager users, you'll be thrilled to know that Google Optimize will automatically import your data layer variables into your experiment to allow you to use for targeting. Any server information that you've stored on the data layer will now be available to help you target a population or exclude a certain group.
Raise your hand if you're ready to be blown away. Let's talk about the advanced targeting options.
If you're a marketer, don't let the word advanced scare you. As it relates to Google Optimize, the word advanced is synonymous with control. Control over the who, when, and where of your experiment.
Most testing tools focus targeting on the present - what page is a person on now or in what city they are currently located. Google Optimize 360 extends these traditional targeting options by allowing marketers import Google Analytics audiences, opening up our targeting to focus not just on the present but the past as well.
You may already be familiar with audiences that you can build and use with Google AdWords, grouping users by their behavior or any custom information we've collected. We can take existing audiences from Google Analytics and duplicate them to use with Google Optimize. This means we can target users that have at one time expressed interest in something or exclude users that have already converted, even if that happened in a different sessions weeks or months ago.
For example, LunaMetrics drives a lot of traffic to our blog, far more traffic than our training pages receive. Using GA audiences we can create an experiment that targets only users on the blog, who have visited a training page, but have not registered for a training. This example combines the present, what page they are on, with the past, they have expressed interest in our trainings and they are not already customers.
Here's how those targeting options would look in Google Optimize.
Not only does this ensure that the right people are being targeted, but it removes people from the experiment as they complete our experiment objective, a training registration (which is a goal we already have established in GA). Taking it a step further, we could use audiences to identify people who saw the experiment and converted, and target them with another test. And all of that is done using Google Analytics data.
Google Optimize has an easy to understand reporting interface that shows you how each variation is performing against the original, and how each variation is contributing to your Google Analytics goal completions. Here's what it looks like.
The results shown above are for an experiment we ran to see if displaying our blog subscription form in different ways would improve the number of people subscribing to our blog (spoiler alert, it did).
Reporting is not unique to Google Optimize, but rather how it collects data and makes it accessible. Unlike other A/B testing tools, Google Optimize does not collect data, it fetches data from the Google Analytics view that is tied to your experiment. That means that the results of your experiment are subject to any filters you have applied to that view. This helps to ensure that unwanted traffic, like internal traffic or bots, does not pollute the results of your experiment and adds another layer of confidence to your experiment results.
That connection with GA also allows Google Optimize to pass experiment results back into the Google Analytics reporting interface. That data is then accessible via the Experiments report (navigate to Behavior > Experiments), where you'll see the same data (experiment sessions, conversions, conversion rate, improvement over the original). And because this data originated from Google Analytics, you are also able to see additional metrics for each variant as well.
* those with a keen eye will notice that the “compare to original" numbers do not match up, more on that here.
But wait, there's more. You can also use Experiment Name, Experiment ID, and variant as secondary dimensions in all the standard Google Analytics reports or you can use the experiment dimensions to create a custom report. The gif below shows a custom report we made that is filtered to include only our blog experiment (using Experiment Name) so we can see how the experiment performed on a page by page basis. Those with Google Analytics access can view the results of the experiment right in the Google Analytics interface, a platform they're already familiar with and that they've been using for years. Now that's a real high-five!
We finish an experiment and the results are conclusive. Now what? For many marketers, to-dos are added to development queues, weeks turn into months, and the next round of experimentation is launched. It is slow. It is labor intensive.
This is where Google Optimize excels. Traffic is simple to allocate between variants. So you can easily shift all of your traffic to see the winning variant going forward while negotiating the logistics of scaling your recommendations.
Chaining experiments together happens quickly. Before long, what started as traditional A/B testing begins to look more like personalization. Users can convert or self-select their way into unique subsets of the audience and have an experience tailored to the audience when the return.
For this website, Online Behavior, it might mean prioritizing new articles based on previous consumption patterns. Users with an affinity for analytics are exposed to more content from the Analytics & Optimization section. Or perhaps audiences are built around the website's personas.
Or career experience. Or degree of technical sophistication. The possibilities are endless.
Our experiments help us learn more about our users, insights that we can easily hand off to Google AdWords or Doubleclick through audiences connected in Google Analytics, keeping our message consistent across our site and other platforms where they'll see our company.
For me, this has been the most impactful result of using Google Optimize. Not only am I learning constantly through experimentation, but through the integrations with other Google products, I can take those lessons learned and improve the experience of my users.
If you liked this post, you might also enjoy Krista Seiden's Building a Culture of Testing and Optimization