In the world of paid social, we’re always being told by platform reps that “creative is the new targeting”, and that creative diversification and volume are becoming increasingly important as platforms lean into AI-heavy targeting. As performance marketers, what this means for us is that we need an ironclad creative testing approach to maximize the learnings we get from platforms, as our insight into targeting continues to devolve. Let’s dive into what this approach looks like.
The Hypothesis
Every great test must start with a clear hypothesis. Doing so keeps the vision for the test clear, prevents random testing, and ensures insights accumulate over time. Here’s an example of what that might look like: “If we use UGC-style video instead of product demos, CTR will increase because the creative appears more authentic.”
Set Your Testing Criteria
To ensure validity, you should plan on all creative tests following the minimum benchmarks below:
50 conversion events per test
Minimum 2-week test window (1 week to get out of learning phase, and a second week to ensure clear results)
Equal budget distribution
If the results aren’t statistically significant, extend the test rather than drawing inconclusive results.
Define Your Testing Variables
For clear results, only one variable should be tested at a time; Isolating variables helps identify what is really driving performance. Here’s an example of variables you might test:
Leverage New Testing Features
Utilize Meta’s creative testing feature to test creatives against one another and designate a specific amount of your budget to the test ads, without having to scale ads in a separate campaign and constantly restart the learning phase.
Structure Your Ad Variations
Creative used in your test should follow a structured format that allows clear variables to be identified and tested. If you wanted to test 6 ad variations for example, it might look like this:
3 Angles: Problem, Benefit, Testimonial
2 Formats: UGC, Product Demo
1 CTA: Shop Now
Follow a Phased Testing Process
Your creative testing should include three main phases: exploration, validation, and scaling.
Exploration: Your primary goal here is to discover what works. Test several different creatives variations (5-10) with different hooks and angles, reviewing engagement metrics like CTR, CPMs, Thumpstop rate, etc.
Validation: Take your top 2-3 winners from phase 1. Test these against existing winners, or other variations of the same hook. Metrics to evaluate are CPA, CVR, ROAS.
Scaling: Once winners are determined, you should increase investment, expand formats and create iterations (shorter version, different hook, new opening frame, etc.).
Utilize a Creative Testing Matrix
Now that you’ve done all this testing, it's important that you store results in a way that’s clear, organized, and easy to reference. By building and updating a creative testing matrix, you can begin to see trends over time of what has worked and what hasn’t. For example, your matrix may look something like this:
There’s no exact science to the perfect matrix, but it should clearly track everything you are testing and provide valuable insights that are easy for you to reference. Were there nuances from your test you should be aware of? Perhaps a creative performed really strongly with one audience, and struggled with another. You matrix should include data that makes it easy for you to report results and iterate going forward.
Identify Winners & Iterate
So you’ve completed your test, filled out your matrix with data, and now you can determine winners and iterate on those ads. This cycle would follow this general pattern: launch new creatives, analyze performance, iterate winners, introduce new concepts, then rinse and repeat.
Ads platforms today are increasingly creative-driven. Formulating a systematic testing plan helps us determine trends and compound insights over time, rather than relying on random experimentation and putting our learnings at the mercy of the algorithm. Let Revel Marketing Partners help you jumpstart your creative testing strategy today.