A Shopify supplement brand had been at $18,000/month in Meta spend for 4 months. Every time they pushed budget above $22,000, ROAS dropped from 2.8 to below 2.4 within a week. They'd tried 3 times. The result was the same each time: efficiency declined, they pulled back, ROAS recovered, and they were back where they started. The problem wasn't their audiences, their bidding strategy, or their creative. It was that the system holding ROAS at $18K wasn't built to hold it at $22K.
What actually breaks when you scale
When a Meta campaign converts efficiently at a given spend level, it's because the algorithm has found a pocket of buyers — a specific combination of audience characteristics, placement, time of day, and creative signal — where purchase intent aligns with what you're selling at the cost you can afford.
Scale the budget by 20%, and Meta has to find 20% more buyers from that same population. In a defined audience, that means showing ads to people who are slightly less likely to convert — fringe buyers, not core buyers. CPMs rise (higher competition for the best placements), and conversion rate on the incremental traffic drops. ROAS falls not because anything broke, but because you've gone past the efficient frontier of your current audience and creative setup.
The pattern is predictable: strong ROAS at current spend → budget increase → 3-5 day learning phase adjustment → efficiency decline → ROAS drop → panic → budget pulled back → ROAS recovers. Brands stuck in this loop aren't doing anything wrong. They're hitting the ceiling of a system that wasn't built for the budget they want to run.
The ROAS decay pattern and what causes it
Three things cause systematic ROAS decay at scale, in order of impact:
1. Creative frequency running ahead of audience size. When you scale budget without adding audience, the same buyers see your ad more often. Frequency climbs. Buyers who've seen the ad 5+ times have already decided. Spend keeps flowing toward a fatigued audience. The supplement brand's top creative had 7.2 lifetime frequency on their warm prospecting audiences when they tried their third scale attempt. Frequency above 5 on any meaningful audience segment almost always predicts a ROAS floor.
2. Landing page CVR not matching the traffic quality. At higher budgets, Meta reaches slightly less qualified buyers. If landing page CVR is already marginal (below 2.8%), it can't absorb the quality drop. A page that converts core buyers at 3.1% might convert fringe buyers at 1.6%. ROAS at scale reflects that gap. Check landing page CVR specifically from paid social traffic — not overall site CVR — before scaling. See the above-the-fold rewrite for what changes CVR fastest.
3. Single-creative fragility. Most brands scaling on one or two top performers find that budget increase accelerates fatigue. Meta routes the extra spend through the highest-performing historical creative. Fatigue arrives faster. Without a creative fallback, there's nowhere to shift spend when the primary creator exhausts.
What the supplement brand changed in 3 weeks
Three changes before attempting the fourth scale:
Week 1: Paused the top creative (7.2 frequency), pushed $40/day behind 3 under-tested assets that had been running at minimal spend. Two of them outperformed the original on thumb-stop rate (32.1% and 29.4% vs. the original's 28.6%) within 5 days. The creative library had depth — it just hadn't been tested properly.
Week 2: Rebuilt the landing page above the fold. CVR from paid social had been 2.4%. The page had a generic hero image, the brand name prominent, and three different CTAs. Changed to: headline matching the specific ad angle ("energy that doesn't crash"), lifestyle photo showing product in use, one CTA. CVR moved to 3.6% within 8 days.
Week 3: Expanded audience from Lookalike 1-3% to Lookalike 1-5% on the winning creative, giving Meta more population to work with at scale.
Fourth scale attempt: $18K → $24K. ROAS held at 2.7 for 3 consecutive weeks. Profitable above their 2.3 break-even threshold throughout.
The test to run before your next budget increase
Before any budget increase above 30% in a 2-week period, answer three questions:
- What is landing page CVR from paid social traffic in the last 14 days? If below 3.0%, don't scale — you're amplifying a conversion problem. Pull this from Shopify UTM data, not Meta.
- What is the lifetime frequency on your top-spend creative? If above 5.0 on any warm audience, you need a creative refresh before scaling. More budget through a fatigued creative accelerates the decay.
- Do you have at least 2 creatives with stable ROAS above break-even for 14+ consecutive days? If only one, a budget increase puts all of your spend behind one asset. When it fatigues — and it will — you have no fallback.
All three yes: increase by 30–50%, hold for 2 weeks. Any one no: fix that constraint first. The budget can wait. The system can't be skipped.