Targeting is no longer the primary lever in paid advertising. Meta and Google's algorithms have become so good at reading user behaviour that manual interest layering is mostly obsolete. If you sell yoga mats, targeting "people interested in yoga mats" is often less effective than targeting a broad, open audience and letting the algorithm analyze who interacts with your creative.
This means your **creative** is doing the targeting. The visual hook, the headline, and the core angle decide who stops scrolling and who clicks. Yet, most brands run creative testing in a way that produces zero clean data.
The Multi-Variable Mistake
The standard testing mistake is changing too many things at once. A marketer launches three ads: Ad A has a founder-led video with a discount headline; Ad B has an influencer testimonial with a benefits headline; Ad C has a static product shot. They run them in different ad sets.
If Ad A wins, what drove the result? Was it the video format? The founder's presence? The discount offer? Or the specific audience in that ad set? You have no idea because you changed four variables at the same time. You can't replicate the win because you don't understand the reason behind it.
The PopLab Testing Framework
To get clean data, we isolate the creative as the only variable. Our framework runs on a simple structure:
- One Control Audience: We use a broad, open audience (or a large, consolidated lookalike) with no restrictive targeting. This is the environment.
- Four Formats, Same Offer: We launch four distinct formats in the same ad set:
- Format 1 (Image): A clean, product-focus static with minimal text.
- Format 2 (Video 1 - Testimonial): A customer-led review addressing objections.
- Format 3 (Carousel): An interactive gallery showing the product in use.
- Format 4 (Video 2 - Emotional/Founder): A direct, founder-led video addressing a core customer pain point.
- Identical CTA & Landing Page: Every ad points to the exact same URL with the exact same offer.
Case Study: The Wellness A/B Test
We ran this exact setup for a women's wellness client in South Africa. The results were stark:
| Creative Format | Website Leads | Cost per Lead (CPL) |
|---|---|---|
| Image (Static) | 79 | $0.52 |
| Video 1 (Testimonial) | 32 | $0.63 |
| Carousel | 44 | $0.52 |
| Video 2 (Winner - Chronic Pain Angle) | 221 | $0.45 |
Because the audience, offer, and landing page were identical, the data gave us a clean read: the emotional, pain-point-driven video (Video 2) was the clear winner. It converted nearly three times as many leads as the runner-up, at almost half the CPL.
We immediately shut down the underperforming formats, moved the winning video to our primary scaling campaign, and scaled the budget. That is how you turn creative from guesswork into a predictable growth engine.
Quit guessing which creative is better. Run clean tests, read the data, and scale what works.