Imagine the following scenario: you are a shoe manufacturer looking to launch a new line of streetwear products. Marketing ideas are aplenty, from catchy messages to powerful creatives and multiple formats. After weeks of planning, the launch date is upon you: you activate the campaign, finger quivering on the mouse, and…
It’s a success.
You’re just not sure why.
Perhaps it was the ad copy that highlighted the killer prices. Perhaps it was that creative of the model jumping with the shoes on. Or perhaps it was the YouTube unskippable ad that brought them together. Or was it the bumper ad?
Testing allows marketers to isolate variables and understand the actual factors that drive performance in order to know which levers to pull to get the most out of their campaigns. One clear advantage of paid campaigns over organic ones is that they allow an advertiser to easily test what works and what doesn’t. Most interfaces allow for campaign test setups, and due to the nature of paid media, results are quick to arrive.
Types of testing in performance marketing
1. Period-on-Period (Time-Based) Tests
In period-on-period testing, you compare performance across different time periods. For example, you may want to compare how your campaigns performed when using Google and Facebook (before) versus how they performed after adding YouTube (after, such as in the following month).
Advantages:
Easy to Set Up: Time-based tests are simple and don’t require complex systems or advanced strategies.
Quick to Implement: They can be run quickly, making them ideal for testing big changes or additions to your strategy without overcomplicating the setup.
Disadvantages:
External Variables: Time-based tests are highly susceptible to external factors. For example, if one period includes a sale event like Black Friday and the other doesn’t, results can be skewed. Seasonal changes, economic shifts, or marketing promotions can also distort your conclusions. This lack of accuracy means that it can be challenging to confidently attribute success or failure to a specific variable, given that so many other factors could impact the results.
2. Geo-Targeted (Geographic) Tests
In geo-targeted tests, you split your campaigns by location to isolate the impact of a particular change. The shoe manufacturer in that scenario could select two comparable cities in terms of performance (RoAS, conversions), run the original campaign in one city (“control”) and run the campaign with the changed variable in the other city (“test”).
Geo-tests are often used by marketers to prove the value of a new brand awareness initiative. For example, if the test city with Display ads running shows a +10% increase in overall sales compared to the control city, it may be that programmatic advertising replenished your prospecting audience and thus your pool of potential customers.
Advantages:
Minimises External Variables: By testing in different geographic locations during the same period, you reduce the risk of external factors (e.g., seasonal changes) skewing the results.
Clear Comparison: You can run different strategies side by side, making it easier to attribute performance differences to the test variable.
Disadvantages:
Regional Bias: Different regions may have varying consumer behaviour, making it difficult to generalise results to the broader audience.
Limited Audience Size: Depending on your product or service, finding comparable regions may be difficult, especially if your target audience isn’t spread evenly across locations.
3. Conversion Lift and Brand Lift Tests
Conversion lift and brand lift tests are more sophisticated approaches that help determine how exposure to an ad or campaign has impacted conversions or brand perception compared to a control group that was not exposed to the ads.
These tests require assistance from the advertising platform and are the most time-consuming to set up, but they are seen as the most accurate measure in the industry.
Advantages:
Precise Measurement: These tests are the most accurate measure of the incremental impact of your advertising efforts.
Control Groups: The use of control groups reduces the influence of external variables, helping to isolate the impact of the ads.
Disadvantages:
Complex Setup: These tests can be complicated to set up as they require the assistance of the advertising platform (such as Google or Facebook)
Longer Time Frames: Conversion and brand lift tests usually require longer time periods to gather statistically significant data, which may delay decision-making.
Minimum spend: Often, platforms have a minimum spend requirement to access lift tests. This can be a barrier for smaller advertisers, as many platforms require a significant monthly budget to run these types of tests effectively.
4. A/B Tests (Split Testing)
A/B testing is one of the most commonly used testing methodologies in digital marketing. It involves running two versions of a campaign or ad (A and B) simultaneously to compare their performance. For example, you could test different headlines, creative elements, or landing pages.
Advantages:
Controlled Environment: A/B tests are typically the most controlled type of test and allow you to test specific variables, such as ad copy, images, or bidding strategies, and directly compare their effectiveness.
Clear Results: Because both versions are running concurrently, external factors (such as seasonal trends) are less likely to impact the results, allowing for clearer attribution of performance.
Disadvantages:
Limited to One Variable: A/B testing works best when testing one variable at a time. Trying to test multiple changes (e.g., both ad copy and creative) can complicate the results and make it difficult to determine which element drove the performance change.
Audience Size: For A/B testing to deliver meaningful insights, you need a sufficiently large audience to reach statistical significance.
Copyright: Silkee Digital Limited
What can you test?
With the multitude of methodologies at your disposal, almost everything can be tested in performance marketing.
Ad Assets, landing pages, targeting options
Assets include headlines, ad copies, sitelinks, creatives and CTAs, while targeting options include audiences Social and Programmatic and Keywords for Search. These elements can easily be A/B tested, ideally at a rate of one variable at a time to ensure accuracy.
New platforms and incremental uplift
If you are running Paid Search and are thinking of adding a branding campaign through TikTok or Instagram to enlarge your prospecting audience, be ready to test their conversion uplift through either a Geo-test or a platform-supported conversion lift test.
Brand perception and incremental uplift
Methodologies for conversion and brand uplift tests will vary per advertising platform. For instance, Google will size brand awareness by running surveys on a test group before and after a brand campaign to measure whether users post-campaign are either more aware of the brand and/or have gained a more positive impression of it.
Campaign structure
With the rapid evolution of digital marketing platforms, campaign structures that were once industry standards quickly become outdated. For example, hyper-relevant setups like Single Keyword Ad Groups (SKAGs) are now falling out of fashion in favour of consolidated structures which seek to feed the algorithm with as much data as possible.
How do I set up my campaigns to be test-ready?
Testing budget
It is possible to test new audiences in Social or Display by simply launching them alongside your standard BAU audiences and see how they perform. However, that may lead to your main campaigns cannibalising your test audience’s budget, which would lead to a lengthened test time due to a lack of statistical significance.
A good way of circumventing this issue is by setting aside a testing budget. On TikTok Ads, you could create an ad group for each audience you wish to test and set identical budgets at ad group level to allow for a fair comparison. On Facebook Ads, you could replicate the same process using ad sets.
Testing structure
Certain platforms advocate for a specific structure to allow for better testing and optimization. For example, an example of modern campaign structure on Facebook Ads to split one’s advertising into two campaigns: prospecting, where the goal is to find new audiences to drive to the website, and retargeting, to convert past visitors who are lower down the funnel.
In that example, audience testing would be part of the Prospecting campaign with a dedicated budget per ad set. Winning test audiences would then be shifted into the BAU ad sets and losing audiences would be discarded.
As best practices change all the time, it is always recommended to check with your platform rep as to the latest findings.
Conclusion: Test, Learn, Adapt
In today's rapidly changing digital landscape, new features and campaign structures continuously emerge, and what works today may not work tomorrow. To stay competitive, you must continually test new approaches, learn from the results, and adapt your strategies. Whether you’re introducing a new campaign type like Google’s Performance Max, trying out broad match keywords, or experimenting with new platforms like YouTube or TikTok, testing is the only reliable way to attribute success to specific changes.
If you are interested in implementing a test-and-learn approach to your marketing campaigns to boost your growth, feel free to contact us.
Kommentarer