Website design is never truly complete. If you’re trying to get more conversions out of that new landing page, it may be time to rethink that hero section or consider adding a video. However, you may not be sure what changes will work best.
Even the most seasoned marketing experts can’t trust that their instinctual analysis will always be correct. That’s why every marketer or business owner needs to be aware of how to run an A/B test in Google Analytics.
In this article, you should learn what an A/B test is as well as how you can easily set up various types of split tests with minimal effort.
An A/B test is a process of utilizing two versions of a sales page to determine what yields better results. You will start by fully designing your landing page including the layout, color choices, images, videos, and copy.
Then, you create a second variation of the page generally by modifying the existing format. These changes depend upon what exactly you are looking to test during this process. For example, you may have two strong approaches to sales copy, but are unsure of which will best speak to your audience. Another example may simply be recording and adding a personal video to engage new prospects.
After you create your two variations, you can direct a set percentage of your traffic to either page. A 50/50 split will see half of the users arriving at Page A with the other half getting the Page B experience.
Google Analytics then gathers various metrics in real-time to help you determine which page is the best performer. These insights may also help you better understand what creatives your audience responds to, so that future marketing ventures are naturally tailored for success.
Any website owner or business owner should be utilizing A/B testing in Google Analytics. Whether it’s a website or a fancy new landing page, you’re making that investment to get you qualified leads and improve your sales. The best way to do that is to constantly analyze what’s getting results and making the necessary improvements along the way.
While the process of setting up A/B tests is relatively simple, it may not be so for newcomers who are unsure of what creatives need testing. As a result, studies show that most of these A/B tests will fail.
The good news is that we generally understand the reasons why most of these end in failure. Before we get into the specifics of how to set up your A/B test in Google Analytics, let’s take a moment to educate ourselves on what not to do when performing your future tests.
Knowing what not to do is just as important as knowing best practices for success. Here is a quick list of things you need to be aware of and avoid to ensure that you make the most of your time and marketing budget:
Making small copy changes may seem appealing (and can yield results when done right), but it’s not a substantial change that generally sees long-term differences. While you and your team might be able to point out the difference in a header, a paragraph, or an image, the overall layout of both variations remains fairly identical.
So – what should you change? Consider big picture items such as the overall tone or approach to selling your product on the page. An easy example might look like this:
In this particular example, Page B is going to outperform Page A most of the time. A quality video will yield better results than still images 75% of the time. Without that insider knowledge, you would need to rely on an A/B test to know for sure that the video was the deciding factor.
It’s great that you have several ideas on how to approach your webpage. However, incorporating too many changes simultaneously may make it more difficult to isolate exactly which changes bolstered your results.
When creating your experiment, you should make an effort to simplify your hypothesis and put the variations to the test. Then, create a separate experiment down the line to further improve conversions. Don’t view website optimizations necessarily as a sprint, but plan on making continuous improvements and revisions over time.
Testing is a continuous process. If you go into the expectation that you’re going to make some quick changes and watch the leads roll in, you’re going to be disappointed.
A/B tests are most effective when you allow sufficient time and conversion volume. It’s common sense that you would put more faith in a survey of 1000 unique persons versus 100. Unless your brand is already a hot-ticket item, it can take several weeks to gather enough data of any real statistical significance.
This one can be a bit harder to identify, but your personal bias may be blocking from you accurately assessing problems with the page. You keep making changes to the areas that should be driving conversions, but neither variation is outperforming the other in any noteworthy fashion.
True testing needs to be focused on what is actually happening, not your personal feelings or assumptions about the copy or web design. Unless you’re confident that you know what to test, A/B testing may not be helping as your variations are not featuring any actionable changes.
To ensure that you prepare properly, let’s take a look at some of the key elements on the webpage that you generally want to be the focus of your A/B testing:
This is a broad category, but debatably the most important part of the page. Every person who clicks on the ad needs to be able to easily read and comprehend what they’re being told on the page.
The description of the product, service, and offer should be clear and easy to digest. It’s okay for your audience to have some questions, but they should feel informed and empowered to sign up or make a purchasing decision on the spot. If you’re getting traffic to the page but have a low conversion rate, explore different ways to communicate the details such as bullet lists, product or service comparisons, and pricing details upfront.
Headlines are the first tool you use on a web page to get your audience engaged. It should directly address the individual and provide them with an immediate incentive to take action and learn more.
We can use another example to better understand an A/B test of two headlines from two landscaping services:
Both headlines directly address the customer by using some form of “you,” but Page B is more engaging. The headline immediately describes the service, provides an attractive offer, and incentivizes immediate action by revealing a time limit. While improving your home sounds like a great benefit, A/B testing allows you to take a step back and determine the actual highlight of the page.
Customers continue to get more purchase-savvy online and tend to be more skeptical of a great offer. You can combat this effectively by offering your customers social proof. Examples of this may be testimonials, past sales figures, or current subscriber numbers.
You then place social proof on your landing page, but you’re still not converting. Thankfully, A/B testing allows us to explore the possible reasons. The metrics may show that users are not scrolling down to see your testimonials at the bottom of the page, meaning that your B variation should try alternative placements to improve the impact and visibility.
This may seem like a no-brainer until you consider how many visually unattractive pages you see every month as an active internet user. Poor color choices, illegible or hard-to-read text and lack of visual engagement with high-resolution images or video can cause users to bounce off your page quickly.
While problems with your web design may not be so severe, you still want to explore alternate page layouts and designs to activate your readers. The design should amplify the copy, gracefully walking readers through your pitch and naturally leading them to a call-to-action button.
After you review the key elements of your webpage and come up with a few actionable hypotheses, you can begin setting up your A/B test. You will require a Google Analytics and Google Optimize account to begin.
After logging into your Optimize account, complete the following steps:
1. Click on the appropriate Container name and navigate to the “Experiments” page.
2. Click “Create Experiment” Give your experiment an easy-to-identify name that’s relevant to the subject of testing.
3. Enter the URL of the page you wish to test.
4. Select “A/B test” Then, click “Create”
Now that your experiment is active, you should see two tabs on the page. One is the Variants card and the other is your Configuration card.
1. On the bottom right side of the page, click “Create Variant” to begin.
2. Name this variant and click “Add.” You can do this again for as many variants as you need to create.
3. You can see your variants in a list on the page along with the number of changes currently present. Click on a variant anywhere to begin making your desired edits.
4. An editor panel will appear, and you can now make changes to the new page variant. When you are satisfied with the changes, be sure to click “Save.”
5. When all of the changes are in place, click “Done” to complete the variant.
Now that you have a variant to test alongside your original page, you need to set the objectives for your Google Analytics A/B testing experiment. You can do this by performing the following steps:
1. On the objectives tab, select a Google Analytics View. Then, select a Primary Objective.
2. Click “+ Add Objective.”
3. Add a description and hypothesis that describes what you’re looking to test and achieve with this experiment. Then, click “Save.”
4. Next, we will select who to target with the new variant and when by going to the Targeting tab.
5. Under Who, select the split percentage of your audience you want to send to your variant page.
6. You can further establish targeting rules by utilizing the When tab. There are a variety of custom rules you can use to determine when your audience will see the variant. We encourage you to review this in-depth targeting tutorial from Google. Also, be sure to review Google’s policies before continuing with your new targeting rules.
When you are satisfied with the experiment, you can click “Start” to push your variant pages live. Google recommends allowing this experiment to run for at least two weeks or until a variation has a 95% probability to outperform the original.
Performing A/B split tests in Google Analytics is easy to pick up but offers a host of ways to get even more confident results with experience and know-how. You may eventually wish to graduate from general A/B split tests to perform more in-depth content experiments.
Google also provides built-in reporting that you can review at the end of your experiment. Their reporting utilizes Bayesian inference, which is a statistical analysis method that helps to determine the actual likelihood of your variant outperforming the original. Optimize can also advise you on when it can no longer realistically gather relevant data to improve the results.
A/B split tests in Google Analytics offer numerous benefits with almost no downside, including:
If you have a page that you’re ready to test, you can use this information to start testing today. Stop wasting money on vanity traffic numbers, and start turning those leads into conversions.