October 26, 2021
It’s no surprise that we love experimentation. It’s the best way to hold constant dozens (if not hundreds) of variables – like promotions, consumer interest, market conditions, competitive pressures, and so on – to give us the ability to effectively measure changes – big or small – made to a website and understand their impact on behavior and conversion.
While there are many testing SaaS providers out there, for many who are just getting started with a/b testing, they want to try something that’s low or no cost and commitment. Welcome to the stage, Google Optimize. This experimentation platform is the perfect tool for brands just getting started with testing as it’s a completely free, scaled-back but full-feature platform with A/B testing and personalization.
A notable aspect of being “scaled back” is that although Google Optimize has some decent help articles out there, they’re mostly focused on the platform and its features, but not ways to actually use it.
Below we will outline a few areas of the test build phase and some tips and tricks for using Google Optimize for more than just building a/b and multivariate tests but for running an experimentation program.
We have one caveat that we have to get out of the way: Google Analytics is required. Data in Google Optimize is shared from Google Analytics (GA). This means that goals in GA are automatically ported over, which is convenient and makes adding goals to your test an easy click-click-done; however, if you don’t have GA you can’t use Google Optimize - full stop. Furthermore, if your GA instance isn’t set up that well, you’re going to struggle each and every time to build your test goals.
Furthermore, to supercharge your GA instance you will likely need to use Google Tag Manager to help instrument events and other listeners that then get passed into GA as Events, which are then passed into Google Optimize as test-specific objectives.
TIP: Build out your events in GTM and pass into GA for easy, reusable goals in Google Optimize to make setting up goals a few clicks away.
Oftentimes one test will beget another or may spark inspiration for testing in other areas of the site. This is almost guaranteed to result in a lot of tests being named similarly or closely related, which may make sense in the moment but doing look-back analysis or bringing someone else up to speed can leave them having a groundhog’s day moment in your Google Optimize archive.
There is a unique test number associated with each test that you deploy (image below), but it’s a bit obfuscated and, in our experience, seldom used; instead, we opt for creating our own test naming convention to keep track of tests across multiple clients in their respective testing programs.
TIP: Come up with your own test naming and / or numbering system that works for you and your business.
Actually you can select multiple elements in the WYSIWYG editor of Google Optimize and we don’t believe this is documented anywhere (until now that is). If you click on an element, hold the SHIFT key, and click on another related element, the Select Elements helper tool will give a suggestion for related elements (gif below).
TIP: rather than making individual, unit-specific changes that may breakdown or fail to grab all elements in question, easily select multiple related elements directly in the WYSIWYG using the shift key.
The debug tool is great for checking your own work and observing a preview of change(s) on the website with the test targeting and inclusion criteria. This tool is slightly limited in that it's not something you can share – this is only available to those working directly in Optimize, but it can be useful if you’re having trouble identifying why a certain test isn’t firing as expected.
When working with clients, stakeholders, or ecommerce managers, they are always the last ones to preview a test. Typically we share a preview twice in our build process – once when building a test and getting design input / feedback from our team and again when the test build is done, our QA has passed, and we’re looking for final approval. And although usually sufficient, Share Preview can be a bit buggy and if the user navigates to a new page, preview mode will likely fail when they return back to the test page.
Sometimes there are tests that require specific triggers or flows from one page to another and debug and preview can only give so much. In order to see an entire user flow or conditional triggers, you might consider adding some query-based preview parameters and pushing the test live to a limited audience, likely you and your stakeholders, clients, and/or team. Query-based preview can be a way for added troubleshooting, test if goals are firing correctly, and see what a test would look like in the real world without being live for the world.
Query-Based Preview: How To: Under Audience Targeting click “+ Add Rule” and you’ll see a side panel open from the right. Under advanced click on Query parameter and you’ll be directed to a second page of this Audience builder (see image below).
Few things are worse in CRO than running a test and checking back in a week or two later only to realize the targeting was set up with a typo and there’s been 0 visitors and 0 conversions.
We definitely recommend – and have built into our process – regular and frequent check ins with our tests, especially in the beginning. But as data is shared from GA to Google Optimize, there is a delay in reporting so you might have to allow 1-2 days for the first results to show up in Optimize.
TIP: If you’re really worried about whether your test is firing correctly you can verify that your experiment is receiving visits by checking the "active visitors" column in the list of active Experiments.
We hope you found these tips and tricks useful and can improve your testing program within Google Optimize. Know too that a lot of these practices can be utilized in other test platforms as well, so when it’s time to level up from Google Optimize to something more robust and full-featured like AB Tasty or Optimizely, they will continue to be useful.
What other tips do you have for Google Optimize? Send them to firstname.lastname@example.org and let’s nerd out.