What is a good ROAS on Meta ads?

By Sofia Marchetti, co-founder & measurement lead · Published December 9, 2025 · Updated August 14, 2026

The median ecommerce brand earned a 1.88 return on ad spend on Meta over the year ending July 2026, measured across 40,000+ stores by Triple Whale. Blended across all channels, the same population medians 3.68. So "good" starts around there statistically, but the honest answer is arithmetic: a good ROAS is one comfortably above your break-even, and break-even is set by your margin, not by anyone's benchmark.

How do you calculate your break-even ROAS?

Divide 1 by your contribution margin. A brand keeping 60 cents of each revenue dollar after product and fulfillment costs breaks even at 1.67; at a 25% margin, break-even is 4.0, which means the median Meta performance of 1.88 loses money for that brand on every order. We put this table in front of every prospect because roughly half arrive targeting a ROAS number they inherited from a podcast rather than their own unit economics.

Which ROAS are you even looking at?

Three different numbers wear the same name. Platform ROAS is what Ads Manager reports under your attribution window. Incremental ROAS, available as an opt-in Ads Manager setting since April 2025, counts only conversions Meta's holdout models estimate the ads caused, and it is nearly always lower. Blended ROAS is total revenue over total ad spend, the 3.68 median above. An account that quotes only the most flattering of the three is not lying exactly, but it is choosing its mirror.

Our reporting shows all three weekly; the method is described on the process page. In the pet-supplements engagement, deduplicating the event pipeline moved reported ROAS from 3.1 down to 2.6 while real performance improved.That write-up is the cleanest illustration we have of why the reported number deserves an audit before it deserves a bonus.

Why did ROAS get harder to read after 2021?

Apple's App Tracking Transparency collapsed the share of iOS users Meta could observe directly; opt-in rates sat near 25 to 37% globally in 2025. Meta modeled its way around part of the gap, which keeps the reported number useful for optimization and fuzzy for accounting. That is the entire case for running a second ledger: your revenue data does not model anything, it just counts.

What improves ROAS in practice?

In the order we deploy them: fix the event pipeline, since delivery steered by clean data simply aims better (Meta's published figure for adding the Conversions API is about 13% lower cost per action). Consolidate structure so budget clears the learning phase; our accounts median 24 days to a stable CPA after restructure. Then feed creative volume, because Meta attributes about 56% of auction outcomes to creative quality. Chasing ROAS through bid tweaks and audience hacks, the 2019 playbook, now ranks a distant fourth.

How does Meta's share of your budget change the math?

Heavily. Ecommerce brands in the Triple Whale panel routed 66.88% of their total ad spend to Meta over the year to July 2026. At that concentration, the gap between Meta-attributed ROAS (median 1.88) and blended ROAS (median 3.68) is mostly a statement about how much revenue arrives through channels Meta cannot see: email, organic, direct, retail. Neither number is a lie. The blended figure includes revenue Meta helped cause but cannot claim; the platform figure claims some revenue it did not cause.

The failure mode is managing to whichever number flatters this month. Pick the pair, report both, and hold the ratio steady quarter over quarter; a drifting ratio is your earliest warning that attribution and reality are separating.

What trajectory is realistic after fixing an account?

Slower than the case-study economy suggests. Across our accounts, the median engagement takes 24 days after restructure just to reach a stable CPA, and the median 90-day CPA improvement is 27%. Translated to ROAS at constant prices, that is roughly a 1.4 moving to 1.9 over a quarter: meaningful, compounding, and nothing like the overnight 4x screenshots. Brands with clean events, strong margins, and creative volume beat that pace; brands starting from a polluted pipeline spend their first month just establishing what is true.

When should you distrust a rising ROAS?

Three tells. Revenue is flat while reported ROAS climbs: attribution is claiming organic sales. The attribution window changed recently: the yardstick moved, not the performance. Retargeting spend share grew: you are increasingly paying to intercept people already on their way to buy. Any of the three is grounds for the incrementality check before scaling another dollar.

The inverse also deserves saying: a falling ROAS is sometimes the account getting healthier. Deduplicating events, tightening a window, or switching to incremental reporting all push the number down while pushing its truthfulness up. Judge the trend only after the measurement is settled, which is why every engagement here fixes counting before touching a single campaign.