When a campaign underperforms, most marketers reach for the targeting dial first. In our experience, creative is responsible for the majority of performance variance — yet it's usually the last thing tested systematically.
Our framework tests one variable at a time across a small batch: five hooks against the same offer, then the winning hook against five different visual treatments, then the winning combination against five different calls to action. Each round runs on a fixed, modest budget until statistical confidence is reached — usually 3-5 days.
The key discipline is resisting the urge to change multiple variables at once. A test that changes the hook, the visual and the offer simultaneously tells you nothing about why it won or lost. Isolate the variable, and you build a genuine library of what works for your audience.
Applied consistently, this is what took one client's ROAS from 1.4x to 4.6x over a single quarter — not a new targeting trick, just a disciplined creative testing loop, run every single week.
