There's one certainty when running paid media accounts for your eCommerce store, that you will have to test. A lot. It almost feels like the testing will never end. Not only is it time-consuming, but also because a hearty chunk of your money will be allocated towards planning, analyzing, and iterating on tests, it's important to understand exactly how to test your ads.
This blog will dive into the art of ad testing for both Google and Facebook accounts. It'll cover some of the best practices, the do's and don'ts of running an effective test, and some popular test ideas that you could start implementing right away.
Ad testing is the process of pairing different ad elements in your account to see how they perform. The goal of ad testing is to determine which elements, such as messaging, design, placement, etc. resonate with the target audience and drive conversions. The closest you'll get to the perfect ad will be through testing different ad concepts.
The importance of ad testing in digital marketing lies in your ability to understand, and iterate, what your customers want to see (or even who your customers are). It allows businesses to measure the effectiveness of their ads and determine which versions are performing their best.
Ad testing is a data-driven method of identifying the messages and types of ads that resonate with your target audience. They can also help you optimize your campaigns for maximum reach and impact, ultimately getting the best ROI.
It's easy to talk about all the different tests that you can run on both Google and Facebook, but it wouldn't be of any value without explaining how to go about conducting the test. Here are the five basic things you need to think about before, during, and after your test.
Testing ads on Facebook is relatively straightforward. There are many different variables that you can test, I dive deeper into some in this section, and Facebook ads manager gives you the tools you need to easily create a test.
One tool, in particular, is the split testing tool, or the A/B testing tool. A split test allows you to isolate a single variable and carry out a controlled test on the platform. Although it’s a very easy tool to use, you can still gather learnings outside of the tool. It can also be quite expensive to carry out the test using the A/B testing tool. It’s good to have a play with it for tests such as changing the destination URL or the CTA, but in some cases, like with creative testing or audience testing, it’s not as valuable.
Testing creatives is a type of test that you will always have to run in your account. This includes testing different creative types, such as static images, videos, and UGC creatives. It also includes pairing different hooks and different elements of your creatives to see what your customers are most drawn to.
There are so many different ways to test creatives. You can do it inside the same campaign, scaling the spend of the specific ad set that reaches your desired goals. Or, you can even create a “testing” campaign, and a “scaling” campaign, and graduate any winning creatives into the scaling campaign, set at a campaign budget optimization (CBO). If you go with the latter, make sure you keep the creative running in the testing campaign until it also takes off in the scaling campaign. You can’t be sure that the creatives will perform the same way.
If you want to have better insight to creative performance, use a tool like the Triple Whale Creative Cockpit. This will let you get more granular with your creative data, view creative performance in digestible visuals, and ultimately make better decisions on which creatives to scale and which to cut off.
Testing different ad copy for your ads can be impactful, but it’s frankly not the most important lever in your ad campaign. I always like to pair the same creatives with different ad copy, to see if any sort of messaging sticks. This can include tests such as using emojis in your ad copy, writing lists or keeping it short and sweet.
Here is the outline of ads within an ad set that I like to use
This way, not only am I testing which ad copy resonates with my audience the most, but I’m also understanding if there’s a specific winning combination of ad copy and creatives that we could use to scale.
Testing different audience variants is always a very impactful test that can help you find the audience that resonates the most with your ads. Before the iOS 14 update, dividing your ad sets into different interests, and testing different interests could help you find the people who wanted to see your ads.
Since the privacy update, this is no longer the way to go. Facebook does not have the same data on its users as it once did, and it can't accurately find who falls under specific interests. With that being said, you can still test audiences, it'll just look different.
Here are the different audiences that you can test at the ad set level:
The amount of tests that you can run on Facebook is a pretty long list. Other tests that you can try are call-to-action tests, where you change the button from “shop now” to “learn more,” or to any other CTA that you think might be impactful. Or, you can even test sending your ads to a specific landing page, the product page, or your homepage. Or you can even test different ad placements.
Just like Facebook, there are many different tests that you can carry out on Google. Since Google shows ads on many different placements (search, display, YouTube, and more) there is so much you can do to optimize each of your campaigns.
For the sake of simplicity, I will focus this section on search campaigns, since they are the first campaign any advertiser on Google will gravitate towards.
Google makes it extremely easy to create a campaign experiment that automatically splits your campaign’s ad spend into two, for the different variations you want to test. The experiments tool on Google is much more valuable than its equivalent on Facebook (the A/B testing tool), and it’s particularly useful when you are testing different bidding strategies.
For a lot of the other tests, though, such as keyword tests, which you are going to want to always be testing, and ad copy tests, which you want to regularly test, then the experiments tool makes less sense to use. It’s crucial to evaluate the test that you want to carry out and understand what tools Google’s platform gives you to carry it out.
Testing different bidding strategies is a test that every campaign will go through at least once. Particularly new campaigns, when you switch them from Manual CPC to any automated bidding strategy. Getting results from changing bidding strategies takes time since Google’s machine needs to learn about your campaign, and the users most likely to convert (so that they can bid accordingly in an auction).
To carry this test out, create an experiment in the experiments campaign, and split the spend 50% between the base and the new campaign.
You should always be testing keywords in your search campaigns. When you set your campaign up with phrase-match and exact-match keywords, you’re testing how they work and how much money they spend. When a keyword spends too much but converts pretty low, that’s when you begin decreasing its bid (if you’re on Manual CPC) or pausing the keyword altogether.
When you feel like your ad group has learned more about what you want to achieve, and you want more volume, then you can begin adding broad-match keywords.
You should also regularly be adding new keywords (that have volume) to the ad group. Google has a valuable tool to do this, and it’s always a great optimization tool.
Testing ad copy is something that has gotten more complicated with the depreciation of Expanded Text Ads. Before, it was quite easy to switch out only one headline and test out what specific message resonated with searchers.
Responsive Search Ads are made to automatically test your different headlines, so it’s a different approach to testing ad copy now. With RSAs, you want to be able to test different ad copy, but you’ll be playing with overall messaging instead. For example, if you run a skincare brand, you can try an RSA about its natural ingredients, and an RSA about its sustainable packaging. Whichever ad performs best will be an indicator of what your customers are most drawn to about your brand.
You can also play with pinning, by pinning different headlines in the same spot, and seeing which one gets shown the most.
You have to be more creative with your messaging now that we are limited to RSAs only.
Just like paid social, the biggest impact that you can have over your Performance Max campaigns are your creatives. Once you set up your campaign, Google will sift through all the different assets that you provide, and will give you a score. This score can range from “Best,” “Good,” or “Low.” It’s your job to keep track of the poor-performing creatives, and constantly switch them out to try better-performing creatives.
A tip here is to use your best-performing creatives on your paid social ads directly in your Performance Max campaign.
Another important test that you can, and should do, is landing page tests for your ads. There are two ways you can go about doing this. The first way is changing the final URL of your ads to different pages. You can test out using either your landing page, the product page, or even your homepage, and see which one your visitors resonate with the most. This method can have a lot of impacts, with very low effort. And you usually wouldn’t need a CRO agency to help you carry out tests.
As for actually changing the landing page itself, there are a ton of landing page-specific tests that you can do. For one, you can do an A/B test, as well as a multivariate test. An A/B test segments a particular quality of the page, and iterates on it. For example, the “Shop Now” button color. If you’re only testing a single feature, either its color or its text, then that’s an A/B test. On the other hand, you can also carry out a multivariate test, where you create multiple versions of the variable you want to test to optimize the page’s performance. Your CRO agency will usually handle most of these, but it’s also good to know in case you want to go ahead and test particular pieces of your website.
On the same lines as changing your actual landing pages to optimize your results, you can also carry out a usability test. Usability testing for landing pages involves evaluating the user experience on a landing page to determine how effectively users can complete a conversion. The goal here is to create the smoothest experience possible for your website visitors.
To ensure the results of your test provide clear and actionable insights, it is recommended to only test one variable at a time. Testing multiple variables at once can lead to confusing and less actionable results. For example, let’s say you create two ads on Facebook, and test different ad creatives and landing pages in both ads simultaneously. If one ad outperforms the other, you won’t know whether it was the ad creative or landing page (or both) that lead to the difference in performance. Testing one variable at a time will allow you to test, iterate, and improve quickly and avoid confusion from your results.
It is recommended to allocate between 10%-15% of your total ad budget towards tests. The remaining 85% - 90% would be used for scaling the winning pieces of your tests. You don’t want to use too much of your total ad budget towards tests, because the ROI during tests is less predictable. That being said, if you allocate too small of your budget to tests, the process of iterating and improving your ads drags out and you’ll be at a higher risk of ad fatigue.
It's important to know what specific metrics you are trying to improve with each test. Although the ultimate goal is always to increase ROAS for eCommerce brands, different tests will have greater impacts on specific stages of the funnel, so it’s important to determine which metrics most accurately reflect the impact of the variable being tested. For example, it may be that you're looking to increase click-through rates, so you'll focus on the ads with the highest click-through rates. When you find a winning ad based on your metric, iterate!
Ad testing is an extremely important part of any marketers job, and knowing how to test effectively can save you lots of time and money. Since in-platform data is notoriously unreliable, using a platform like Creative Cockpit for ad creative testing can let you easily view creative performance data with crystal clarity, and will allow you to make informed, data-driven decisions!
Supercharge your growth with a purpose-built ecomOS for brands and agencies.