Body Sculpting / Weight Management Brand

Setup

A direct-to-consumer body-sculpting brand targeting women with a VSL-driven cellulite and fat-reduction offer came to SkyHouse below breakeven. Early testing produced scattered signals — an "aerobics" angle with strong initiate-checkout rates but no purchases, a "legs" angle that spent without converting — and one profitable creative buried in the noise. SkyHouse restructured the account, isolated the winner, and used it as the foundation to scale the offer from sub-1.0x ROAS to a sustained 1.48–1.60x ROAS in 30 days.

Summary

The offer ran from November 2025 with mixed performance, hitting a sub-1.0x ROAS trough in January–February 2026. SkyHouse identified a single profitable creative — the "butt" ad — running at $54 CPA against a $64 CPA target, with low volume because no one had committed scale budget to it. The February–March restructure consolidated spend around this winner, introduced a Winners Scale CBO structure, and launched a new lander format matched to the winning creative's visual language. By March 2026, ROAS was 1.60x.

Challenge

• Sub-Breakeven ROAS: January–February 2026 returned 0.99x or below on the offer — spending more than it earned every week.
• Winner Hidden in the Noise: The profitable creative existed but was underfunded. Its low volume ($54 CPA vs. $64 target) was obscured by neighboring losing campaigns draining budget.
• Wrong Scale Structure: Scale spend was being pushed through Fernando's control before a clear winner was identified — resulting in good spend on an unsorted signal.
• Lander Mismatch: The long-form lander was built for the old copy set. When Matt's new visual-first winners emerged, they needed a different lander entry point.
• Algorithm Without Time: New campaigns were being killed before Meta's algorithm completed audience learning.

Approach

The intervention had four parts:

• Identify the winner: Rigorous per-creative performance analysis found the profitable creative ($54 CPA vs. $64 target) buried under losing campaigns.
• Consolidate scale spend: Matt's Winners Scale CBO structure ($7,989/day at scale on March 1) committed full budget to the confirmed winner instead of spreading it across 5+ competing angles.
• Switch to ROAS bidding: Shifted from direct CPA bidding to ROAS bid at 1.9x — telling Meta to find higher-value buyers, not just the cheapest conversions.
• Match the lander to the creative: Matt's "LFT5 Image" lander replaced Fernando's long-copy lander — matching the image-forward creative style of the winning ad.

Ongoing testing used a 2/3 + 1/3 split: two-thirds of budget iterating on the winning MG3 campaigns; one-third on net-new angle discovery. July 2026 strategy shift: "Marinate CBO" — run a single CBO with ROAS bid for a full week, letting Meta's algorithm stabilize before evaluating.

What We Learned

The winner was already in the account. The profitable creative had been running at $54 CPA against a $64 target — but it was underfunded, mixed in with losing campaigns, and invisible without rigorous per-creative performance analysis. The structural fix — identifying the winner, consolidating scale spend on it, switching to ROAS bid, and matching the lander to the creative format — took 30 days and recovered the offer from sub-breakeven to 1.60x ROAS. You don't always need new creative. You need to find the winner that's already working and get out of your own way.