Meta is lying about your winning ads, and it is the single biggest reason ecommerce brands cannot scale past their current plateau. Meta ads winning ads incremental attribution is the methodology we use at Skaleit to separate ads that genuinely acquire new customers from ads that are simply harvesting purchases from existing customers and engaged audiences. The standard ROAS (Return on Ad Spend) you see in Ads Manager blends new audience conversions with retention conversions, which makes high-ROAS ads look scalable when they are not. As of 2026, the only reliable way to identify a true winning ad on Meta is to filter by audience segment and then validate with the incremental attribution metric inside Ads Manager.
In this guide we walk through the exact account-level diagnosis we run for our clients, the specific Ads Manager breakdowns to apply, and the structural fixes that stop your scaling campaigns from collapsing the moment you increase spend.
TL;DR
– Meta inflates ROAS on ads that pick up existing-customer purchases, hiding the real new-customer winners. – In one client account, an ad showing 6.89x ROAS dropped to 0.81x incremental attribution on new customers. The actual winner showed 3.81x standard and 2.15x incremental on new audiences. – Fix: in Ads Manager, go to Breakdown > Audience Segments, then Attribution Setting > Incremental. – Keep optimizing campaigns on 7-day click / 1-day view for data volume, but make scaling decisions using incremental attribution. – Always exclude existing customers from acquisition campaigns (30, 60, or 90 days depending on the category).
Why Meta’s Reported Winning Ads Are Often Fake Winners
False ROAS happens when an ad is credited for purchases that would have happened anyway. In a recent client account we audited, the ad receiving the most budget had 34 purchases, $2,000 in spend, and a 2.83x ROAS. Another ad in the same campaign showed a 6.89x ROAS, which on the surface looks like a clear scaling candidate.
When we broke that 6.89x ad down by audience segment, the majority of its purchases came from existing customers, not new audiences. That is not a scalable ad. The moment you duplicate it into a separate scaling campaign and push more budget, the incremental return collapses because there is no new demand to capture. You are paying Meta to re-target buyers who would have purchased organically.
This is why so many brands experience the same pattern: an ad performs well at $50/day, gets promoted to a scaling campaign at $300/day, and the ROAS drops in a week. You did not scale a winner. You scaled a retention ad disguised as an acquisition ad.
How to Find Your Real Winning Ads in Ads Manager
The diagnosis takes two breakdown layers inside Meta Ads Manager. Here is the exact sequence we run.
Step 1: Break Down by Audience Segment
In Ads Manager, click Breakdown > By Delivery > Audience Segments. Meta will split each ad’s conversions into:
– New audience – Engaged audience (still counted as new for acquisition purposes) – Existing customers
Filter for ads where the majority of purchases come from new audience or engaged audience. In the client account above, the ad with 11 purchases at $700 spend and a 2.82x ROAS coming almost entirely from new audiences was a stronger candidate than the 6.89x ad fueled by existing customers.
Step 2: Switch Attribution to Incremental
Go to Breakdown > Attribution Setting > Incremental. This shows Meta’s modeled estimate of conversions that would not have happened without the ad. Meta documents this metric in their official Ads Manager attribution documentation.
The difference is brutal:
| Ad | Standard ROAS | Incremental ROAS (new audience) | |—|—|—| | Ad A (existing-customer heavy) | 6.89x | 0.81x | | Ad B (new audience) | 3.81x | 2.15x |
Ad B is the real winner. Ad A would burn budget the moment you scaled it.
How to Optimize and Scale Without Killing Data Volume
A common question we get: should you just optimize every campaign for incremental attribution at the ad set level? You can, but there is a tradeoff. Optimizing directly for incremental attribution causes Meta to report fewer conversions, which means less data signal feeding the algorithm, which means weaker delivery, especially in the post-Andromeda environment where the system relies on volume to model audiences.
Our approach at Skaleit:
- Keep optimizing campaigns on 7-day click / 1-day view so Meta has enough conversion data to learn.
- Make scaling and killing decisions using the incremental attribution column. This is a manual decision layer, not an optimization signal.
- Run a small percentage of budget on incremental-optimized ad sets as a control group if you have the spend volume.
This hybrid lets the algorithm operate on dense data while you protect the account from scaling ghost winners.
Exclude Existing Customers From Acquisition Campaigns
The second structural fix is exclusions. If you are running an acquisition campaign and you have not excluded existing customers, every “winner” you find is contaminated. The exclusion window depends on the category:
– Supplements and consumables: exclude all existing customers from acquisition. Run a separate retention campaign. – One-time purchase products (cold plunges, mattresses, furniture): exclude all existing customers indefinitely. Retention is not the play. – Fashion and apparel: exclude last 30 to 60 days depending on repurchase cycle. – New product launches: exception. You may deliberately target existing customers or engaged audiences to push the upgrade.
For supplement brands specifically, we cover the full acquisition vs retention split in our supplement brand scaling framework.
What This Means for Scaling Decisions in 2026
With Meta’s algorithm update changing how attribution and delivery work, the gap between reported ROAS and incremental ROAS has widened. The platform is increasingly good at finding the easiest conversions, which often means re-touching people already in your funnel.
When we audit accounts that have stalled, the pattern is almost always the same:
– 60 to 80% of “top spend” ads are existing-customer or engaged-audience harvesters. – The genuine new-customer ads are getting underfunded because their standard ROAS looks weaker. – Scaling campaigns are duplicating the wrong creatives.
The fix is not a new creative strategy or a new campaign structure first. It is a measurement fix. Once you can see which ads actually bring new customers, the scaling structure becomes obvious. We pair this diagnosis with the Facebook ad structures we use post-Andromeda to consolidate budget behind real winners.
The Skaleit Workflow for Auditing Winners
Here is the weekly cadence we run on client accounts:
- Pull a 7 to 14 day window in Ads Manager.
- Sort by amount spent, top 10 ads.
- Breakdown by audience segment. Flag any ad where existing customers exceed 30% of purchases.
- Switch attribution to incremental. Note the gap between standard and incremental ROAS.
- Rank ads by incremental ROAS on new audience only.
- Scale the top 2 to 3 by incremental performance, not by standard ROAS.
- Kill or pause ads where incremental ROAS on new audience is below 1x.
This single workflow recovers more performance than any creative refresh in our experience, because it stops you from feeding budget to ads that were never winners.
FAQ
About the Author
Antonio Ventre is the founder of Skaleit, a Meta ads agency that has managed over $10M in ad spend for ecommerce brands across supplements, fashion, swimwear, and DTC consumer goods. Skaleit specializes in post-Andromeda scaling, incremental attribution analysis, and full-funnel creative systems.
Want Skaleit to Build This System for Your Brand?
If your scaling campaigns keep collapsing and you suspect Meta is showing you ghost winners, we run a free audit covering audience breakdowns, incremental attribution, and exclusion structure. Book a call with Skaleit and we will show you exactly which of your top-spend ads are real winners and which are draining budget.