When Meta spends the majority of your budget on the worst performing ad in a campaign, the correct move is not automatic. As of 2026, our rule at Skaleit is simple: if that top-spending ad has a ROAS (Return on Ad Spend) more than 30% below the campaign average AND its frequency is not significantly lower than the average, turn it off. In two client campaigns, applying this rule pushed ROAS from 1.57x to 2.91x in 48 hours and from 3.84x to 4.26x over the following month. This guide breaks down the exact framework, the numbers, and when to keep the ad running instead.
TL;DR
– Meta’s lowest-cost, highest-volume bidding naturally funnels spend to the ad it thinks can scale, not always the ad with the best ROAS. – The 30% rule: if your top-spending ad’s ROAS is more than 30% below the campaign average, turn it off, unless its frequency is more than 30% lower than the average (meaning it’s carrying top-of-funnel distribution). – Case study 1: Turning off a 0.72x ROAS ad lifted account ROAS from 1.57x to 2.91x in two days. – Case study 2: Same fix on a 1.22x ROAS ad lifted campaign ROAS from 3.84x to 4.26x over 30 days on higher spend. – Frequency matters because a frequency under 2.0 usually means top-of-funnel reach; killing that ad without a replacement collapses distribution.
Why Meta Spends Your Budget on the Worst Performing Ad
Meta’s delivery system optimizes for lowest cost per purchase AND highest volume, not pure ROAS. When you use the default Highest Volume bid strategy, the algorithm finds a balance between the cheapest conversion and the largest addressable audience. The ad with the biggest scaling ceiling often wins the budget, even if its return is lower than a smaller, more efficient creative running beside it.
This is why gurus tell you never to turn off the top-spending ad. Their logic: that ad is anchoring your top-of-funnel distribution, and killing it will crash the rest. That logic is partially correct. It is also why so many ecommerce brands leave broken ads running for weeks and watch their blended ROAS bleed out.
The truth sits in the middle. Sometimes the top-spending ad is genuinely holding the campaign together. Sometimes it is just eating your budget. The framework below tells you which is which.
Case Study 1: 1.57x to 2.91x ROAS in Two Days
A brand running from 29 September to 6 August had three ads inside a single campaign. The top spender was pulling a 0.72x ROAS while the overall account ROAS sat at 1.57x. On 6 October, we turned that ad off.
– Day of turn-off: account ROAS jumped to 2.50x – 6 to 8 October: overall ROAS climbed to 2.91x – Uplift: +85% blended ROAS in 48 hours
The skeptical objection is always the same. Two days is noise. Fair. Which is why we ran the same experiment on a longer window.
Case Study 2: 3.84x to 4.26x ROAS Over 30 Days
The second campaign held 19 creatives with one ad soaking up most of the budget at a 1.22x ROAS while the campaign average was 3.84x. We turned that single ad off on 29 August. No other changes. Same 19 ads before and after.
– Before (1 to 28 August): $800 spend, 3.84x ROAS – After (30 August to end of September): $994 spend, 4.26x ROAS – Result: +11% ROAS on higher spend, which for this brand hit the 4x target
This is the key point. When the top-spending underperformer is turned off, Meta redistributes budget to the remaining ads. If those ads have real conversion power, blended ROAS climbs. If they don’t, distribution collapses. Your job is to know which scenario you are in before you click pause.
The 30% Rule: When to Turn Off vs Keep On
The Skaleit 30% deviation rule is what we use across most ecommerce accounts to make this call quickly. It compares the underperforming top-spender against the campaign average on two variables: ROAS and frequency.
Turn OFF the ad if:
– ROAS is more than 30% below the campaign average – AND frequency is NOT more than 30% below the average
Keep the ad ON if:
– ROAS is less than 30% below the campaign average – OR frequency is more than 30% below the average (it is carrying top-of-funnel reach)
Frequency is the variable most people ignore. A frequency under 2.0 (typically 1.5 to 1.7) signals the ad is reaching cold, top-of-funnel audiences. A frequency above 2.0 means it is hitting middle and bottom of funnel. If you kill the only ad reaching cold audiences, the rest of your campaign starves. This is straight out of post-Andromeda distribution logic where creative diversification across awareness stages determines whether budget can compound.
Worked Example 1: Turn It Off
Three ads, each spending $300:
| Ad | Spend | ROAS | Frequency | |—|—|—|—| | Ad 1 (top spender) | $300 | 1.3x | 1.5 | | Ad 2 | $300 | 2.5x | 1.6 | | Ad 3 | $300 | 3.0x | 1.8 |
– Campaign average ROAS: 2.09x – Campaign average frequency: 1.63 – Ad 1 ROAS is 37.8% below the average (>30% threshold hit) – Ad 1 frequency is only 6.5% below average (does NOT qualify as top-of-funnel exception)
Decision: turn off Ad 1. The 37% ROAS gap is real underperformance, not a top-of-funnel burden.
Worked Example 2: Keep It On
Same three ads, but Ad 1’s ROAS is now 1.9x:
– Campaign average ROAS: 2.36x – Ad 1 ROAS is 19.5% below the average (below the 30% threshold) – Ad 1 frequency is 6.25% below the average
Decision: keep Ad 1 running. The gap is inside the normal variance band and killing it would just force Meta to relearn distribution for no reason.
When This Framework Breaks (And What to Do)
These thresholds are not universal. Every account has its own target ROAS, its own creative volume, and its own funnel structure. In our agency work we adjust the rule based on:
– Target KPI: a brand chasing 2x will tolerate more variance than a brand at 4x – Creative volume: with only 3 ads live, turning one off is riskier than with 19 – Awareness stage mix: if you are running a creative sequencing strategy across awareness stages, an ad with unique top-of-funnel role should almost never be paused – Attribution reality: some winning ads look bad in Ads Manager but drive blended lift, which is a deeper issue we cover in our incremental attribution breakdown
This decision also lives inside the broader Andromeda-era shift. If you want the full context on how Meta’s algorithm now distributes budget across consolidated CBOs (Campaign Budget Optimization), read our complete guide to Meta’s Andromeda update, which is the pillar this framework fits inside.
Meta’s own documentation on bid strategies confirms the lowest-cost logic, and Meta’s attribution guidance reinforces why single-ad ROAS in Ads Manager can mislead blended performance decisions.
FAQ
About the Author
Antonio Ventre is the founder of Skaleit, a Meta ads agency scaling ecommerce brands past 7-figures with over $10M in managed ad spend across fashion, supplements, skincare, and DTC verticals.
Want Skaleit to Build This System for Your Brand?
If Meta is dumping your budget into underperformers and you want an expert audit of which ads to kill, keep, or scale, our team can build the exact decision framework into your account. Book a free strategy call with Skaleit.