I turn audience insight into creative that scales. 🌷

Protect's top performers were simple talking-head ads about insurance. They were profitable — but the brand didn't fully understand why, which made the wins hard to repeat with intent.

I went into the comments instead of the dashboard. Most engagement wasn't about insurance — people were commenting that the presenter was beautiful. Surface-level, but that engagement fed the algorithm, which kept pushing the ad. So it was winning on a weak, accidental signal.

Replace it with a relevant signal tied to the buying decision, and the same mechanism pushes harder and far more durably.

• Led with insurance's most common myth — that shopping around is slow and tedious — to earn attention on a relevant hook.
• Kept that format as the template and rewrote the script to the audience's real objection.
• Productionised it with AI avatars so new messaging could be tested almost instantly.
• For a higher-income bracket, I protected the format and changed only the signalling — a high-end setting + tuned script. It scaled to similar spend.

We struggled to find top-of-funnel winners. The category's default formats — UGC and traditional direct-response — weren't landing with this audience, which capped profitable scale.

Instead of "what ad should we make next," I asked "what does this audience choose to watch when no one is selling to them?" Three things aligned:

• Default formats failed for a reason — this audience is skeptical of anything that feels like an obvious ad.
• Their consumption skewed to story — documentaries, history, narrative.
• A cross-niche brand had proven the mechanism — documentary storytelling could carry a product without a hard sell.

I adapted the documentary style to our product — leading with narrative, weaving the product in. The first test validated fast: profitable and scalable within roughly the first week. Then I rebuilt the core of why it worked into new scripts and storylines, and ported it across SKUs.