CASE STUDY · 2026 · BEAUTY
3 WEEKS · CRO + EXPERIMENTATION

Which creative actually works? +73% CVR, answered.

The Lip Bar ran a wide set of paid-social creatives without a rigorous read on which ones drove conversions. I analyzed the library to isolate the true performers, leaned spend into their winning attributes, and lifted paid-social ad CVR by 73%.
AD CONVERSION RATE
+73%

After leaning into the winners

CREATIVE SCORING METHOD
1

Repeatable, not a one-off

ANALYSIS TECHNIQUES
3

Video, text, and image mining

The problem.

The Lip Bar was running a wide set of paid-social creatives without a rigorous read on which ones actually drove conversions. Spend was spread across ads on gut feel and surface metrics, not on a real understanding of what was working, so budget followed whatever felt good that week.

The approach.

I analyzed the creative library to isolate the true performers, and there's a real toolkit for this beyond eyeballing a dashboard. Video analysis reads scene pacing, hook timing, and on-screen text density. Text mining of captions and hooks surfaces the language that lands. Image analytics on thumbnails picks up dominant color, composition, and framing. Pattern-matching the winners against those features turns "this ad felt good" into "these specific attributes convert."

The outcome.

Leaning spend into the attributes that actually converted lifted paid-social ad CVR by 73%. More valuable than the number: the analysis became a repeatable method, a way to score the next batch of creatives against what wins, instead of relitigating the last batch.

+
ARTIFACTS

From the work.

Guessing which ads to scale?

If you're spreading spend on gut feel, I'll tell you which creative attributes actually convert, and give you a way to keep scoring them. One slot open for Q3 2026.

BOOK A 20-MIN CALL