In short
For Amazon brands, Meta ROAS is usually a credit score under Meta's window, not the sales you would lose if the ads stopped. Meta can't see Amazon checkouts. Amazon Attribution only sees tagged clicks for fourteen days. Long windows and healthy-looking ROAS above break-even are where over-credit hides. A geo holdout is the check that turns ROAS into cause. This reality check scores your channel mix, Attribution tagging, window, holdout history and break-even awareness. A high score means treat Meta ROAS as a suspect until you measure. A low score means your setup is less prone to the Amazon mirage, and you should still verify with a holdout before big budget moves.
What “real” Meta ROAS means for Amazon brands
ROAS is return on ad spend: credited sales divided by spend. Meta's version uses its attribution rules. Those rules are useful for creative debugging. They aren't a proof that Amazon orders would fall by the same amount if you paused the ads.
"Real" here means close to incremental ROAS: sales caused by the ads. For brands that sell on Amazon, the gap between Meta ROAS and incremental ROAS is often the whole story. View-through credit, branded search, Subscribe & Save and untagged paths all scramble the picture. This quiz doesn't invent your lift. It flags when your setup makes flattering Meta ROAS especially likely.
How the reality-check score works
Each answer adds points when the setup usually inflates or obscures Amazon incremental sales:
- Sales channel: mostly Amazon scores highest; Meta can't see those checkouts.
- Amazon Attribution: missing tags leave Meta→Amazon paths unreconciled.
- Attribution window: longer windows credit more always-going-to-buy shoppers.
- Holdout history: never measured means ROAS is still only credit.
- Break-even context: a fat Meta ROAS above break-even is where waste hides.
Rough bands: high scores read as likely overstated, mid scores as mixed, low scores as more credible but still unverified. Pair the quiz with the attribution gap calculator when you have numbers from Ads Manager and Amazon Attribution side by side.
Worked example
The calculator's example is an Amazon-first brand with no Attribution tags, Meta's default window, no holdout, and a Meta ROAS that looks well above break-even.
- Channel, tags, window, holdout and break-even answers each add risk points.
- Total score 13 out of 15 → band "likely overstated."
- Drivers call out Amazon-blind Meta reporting and the missing holdout.
That brand shouldn't scale from Ads Manager alone. Size a test with the brand size holdout calculator, then measure.
What to do with each band
- Likely overstated: freeze big budget changes. Instrument Attribution where you can, learn break-even after fees, and plan a geo holdout.
- Mixed: you have some of the right pieces. Run an attribution-gap check and a powered holdout before you trust Meta ROAS for scale decisions.
- More credible: site-heavy sales and a past holdout help. Re-measure after creative or offer changes; credit still drifts.
None of the bands replace math on your own orders. They only tell you how loud to be about distrusting the dashboard before you change Meta spend.
Common mistakes
- Treating the quiz as a lift number. It scores setup risk, not percent lift.
- Optimizing Meta ROAS after a bad score. You will train the algorithm on credit. Measure cause first.
- Skipping break-even. Without margin after Amazon fees, you can't tell if a "good" ROAS clears profit.
- One holdout forever. Creative, season and offer change the truth. Re-test.
How this connects to true incrementality
Incrementality is the sales caused by the ads. Meta ROAS isn't that number for Amazon. After this check, estimate waste with the ad waste calculator, plan states with the geo holdout states calculator, and read how lift is calculated.
Spillmark runs geo holdouts for Meta ads and measures Amazon orders by ship-to state so ROAS can mean cause, with a range.
Frequently asked questions
Is my Meta ROAS real for Amazon sales?
Usually not as incremental truth. Meta can't see Amazon checkouts, so its ROAS is credit under its window. A geo holdout measures what you would lose if the ads stopped.
Does Amazon Attribution fix Meta ROAS?
It helps reconcile tagged clicks, but it still misses view-through and untagged paths and uses its own 14-day click window. It isn't a holdout.
What score means I should pause spend?
None by itself. A high score means don't scale from Meta ROAS alone. Measure before you cut or raise budget.
Can Shopify brands ignore this?
Site checkouts are easier for Meta to see, so the mirage is smaller, but credit still isn't cause. Holdouts still matter near break-even.
How do I measure real ROAS?
Run a powered geo holdout, divide caused sales by spend for incremental ROAS, and compare that to break-even after fees.