A Facebook Ads ROAS fix is a targeted account cleanup that reallocates budget from underperforming ads to proven winners, without touching creatives, offers, or landing pages. In this case study, we took a new Skaleit client from a 0.40 ROAS (Return on Ad Spend) to 1.44 ROAS in just 7 days, a 261% increase, while also scaling weekly spend from $8K to $14K. This all happened during Black Friday 2025, when CPMs were rising, and we hadn’t even launched their Black Friday campaign yet.
This is the exact framework we run on 9 out of 10 ad accounts we audit, because nearly all of them are making the same mistakes.
TL;DR
– Result: +261% ROAS (0.40 to 1.44) in 7 days on a Meta ad account, without new creatives, offers, or landing pages. – Spend scaled: $8K to $14K weekly (+74%), while CPA dropped from $96 to $25. – CPM headwind: CPMs still rose 26% ($8 to $11 average) due to Black Friday competition, and we still hit the result. – The fix: 5 steps: audit KPIs, kill underperformers 15 to 25% above CPA target, reallocate budget to winners, scale on proven winners only, stop daily budget tweaks. – The core insight: Meta’s algorithm often spends on high-engagement ads, not high-conversion ads. Manual reallocation beats letting CBO decide.
Why 9 Out of 10 Meta Ad Accounts Are Burning Budget
Underperforming ads audit is the first thing we do on every new client account, and the results are almost always the same: a large chunk of daily spend is going to ads that are 15 to 25% (or more) above CPA target. This happens because most brands don’t review their ads daily, and Meta’s algorithm doesn’t self-correct as cleanly as advertisers assume.
A common misconception in 2026, especially post-Andromeda, is that Meta’s algorithm will always route spend to the highest ROAS creative. In reality, CBO (Campaign Budget Optimization) frequently allocates spend based on engagement signals like CTR and thumb-stops, not pure conversion efficiency. That means a high-CTR, low-purchase ad can quietly drain 30 to 40% of your daily budget while a genuine winner sits starved.
According to Meta’s own Ads Manager documentation, CBO distributes budget in real time based on “lowest cost opportunities,” but the model optimizes for the campaign objective within available signals, and with sparse conversion data those decisions get noisy fast.
As of 2026, with post-iOS attribution loss and Andromeda’s ML-driven distribution, human oversight on ad-level performance is more critical, not less.
The 5-Step Facebook Ads ROAS Fix (Exact Framework)
Here is the exact step-by-step we ran on this account between November 20 and November 26, 2025.
Step 1: Analyze Current Ad KPIs at the Ad Level
We pulled all active ads across the account and sorted by CPA (Cost Per Acquisition) and ROAS. Not campaign level. Not ad set level. Ad level. This is where the leaks live.
We flagged every ad currently spending meaningful daily budget and split them into two buckets:
– Winners: CPA at or below target, ROAS above account average. – Losers: CPA 15 to 25%+ above target.
Step 2: Kill the Budget Wasters
The CPA target threshold rule we use: any ad running 15 to 25% above the account’s CPA target gets turned off. No second chances, no “let it run one more day.” On this account, we turned off roughly a third of active ads in the first 48 hours.
That recovered budget doesn’t disappear. It goes to Step 3.
Step 3: Reallocate Budget to Proven Winners
This is the winning ads consolidation step. We identified strong performers that Meta had been under-allocating (often because they had slightly lower CTR but much stronger purchase behavior), pulled them into a dedicated CBO campaign, and increased the budget on that campaign specifically.
The logic: instead of trusting the algorithm to “find” the winners inside a messy campaign structure, we manually built a clean campaign with only proven ads and forced spend into it.
Step 4: Scale the Budget on Winners Only
Once we had a clean pool of winners, we increased total weekly spend from $8K to $14K, a 74% jump. Two reasons this worked:
1. Every new dollar went to proven winners, so scaling was mathematically profitable. 2. More spend = more data = more stable algorithmic performance. Meta needs volume to optimize predictably, and starving winners keeps them stuck in a low-signal state.
This is the same scaling logic we cover in our Meta Ads structure for 2026 post, applied to an existing account with existing creatives.
Step 5: Stop Touching the Ads Every Day
The roller coaster effect is what happens when a brand or media buyer changes budgets, duplicates ads, or turns things on and off every single day. The account looks like a heartbeat monitor: spike, crash, spike, crash.
We check the account daily. We do not change the account daily. There is a big difference. On this client, the previous team was adjusting budgets every 24 to 48 hours, which reset the learning phase repeatedly and created inconsistent delivery. We slowed the change cadence, and performance stabilized within 4 days.
The Numbers: Before vs After
| Metric | Before (pre-Nov 20) | After (7 days) | Change | |—|—|—|—| | Weekly Spend | $8K | $14K | +74% | | ROAS | 0.40 | 1.44 | +261% | | CPA | $96 | $25 | -74% | | CPM | ~$8 | ~$11 | +26% (Black Friday headwind) |
And this happened before we launched their actual Black Friday campaign. The result came purely from account hygiene, not new offers or new creatives.
Why This Works Even Post-Andromeda
Meta’s Andromeda update shifted a lot of weight toward broad targeting and creative diversification, but it did not remove the media buyer’s job. If anything, ad-level analysis matters more because the algorithm distributes spend across a wider pool and can dilute your best assets faster.
The fix we ran here isn’t a targeting fix or a creative fix. It’s a budget reallocation fix, and it works on top of any account setup, whether you’re running consolidated CBO, ASC (Advantage+ Shopping Campaigns), or a hybrid.
For a deeper dive into how to analyze which ads are actually winning versus which ones just look like winners, see our Meta Ads analysis system for 2026.
Common Mistakes That Kill This Fix
– Judging ads by ROAS alone: A single-day ROAS spike doesn’t make an ad a winner. Look at 7-day rolling data with sufficient spend. – Trusting CBO to find winners: CBO optimizes on engagement-adjacent signals. Manually consolidating your proven winners into a clean campaign almost always outperforms. – Scaling before cleaning: Adding budget to a messy account amplifies the waste. Clean first, scale second. – Daily budget tweaks: Resets learning and creates the roller coaster. Give changes at least 3 to 4 days to settle. – Ignoring CPM context: A rising CPM (like Black Friday) will hurt ROAS on the same creatives. Factor seasonality in before you kill an ad.
FAQ
How fast can I see results from a Facebook Ads ROAS fix?
In our experience, meaningful ROAS improvement shows within 4 to 7 days if you kill underperformers aggressively and reallocate budget cleanly. This client hit +261% in 7 days.
What CPA threshold should I use to kill an ad?
We kill any ad running 15 to 25% above the account’s CPA target with sufficient spend to be statistically meaningful (at least 2 to 3x your target CPA in spend).
Do I need new creatives to fix a bad ROAS account?
No. In 9 out of 10 accounts we audit, the ROAS problem is budget allocation, not creative quality. Fix the allocation first. Only bring in new creatives once the account is clean.
Does this work with Advantage+ Shopping Campaigns (ASC)?
Yes, though ASC gives you less ad-level control. The core principle still applies: identify which specific ads inside your ASC are winning, and use ad-level exclusions or fresh ASC campaigns built around proven creative pools.
How often should I change budgets in a Meta ad account?
We review daily but change 2 to 3 times per week maximum, and never twice on the same ad within 72 hours. This avoids the roller coaster effect and lets the algorithm stabilize.
Will this fix work post-Andromeda in 2026?
Yes. Andromeda changed how Meta distributes creative and audience signals, but budget waste on underperforming ads is still the number one leak in ecommerce accounts we audit as of 2026.
About the Author
Antonio Ventre is the founder of Skaleit, a Meta ads agency for ecommerce brands. Skaleit has managed over $10M in ad spend across supplement, fashion, skincare, swimwear, and toy brands, specializing in post-Andromeda scaling strategies.
Want Skaleit to Build This System for Your Brand?
If your ad account is spending on underperformers and you want us to run this exact 5-step fix on your brand, book a free audit with Skaleit. We’ll show you exactly where the budget is leaking before we take on the account.