platform overclaim
Four platforms, one sale. Check whether what they claim is even possible against the revenue you booked.
Use the same period for everything. Revenue is from your own system (store, payments, CRM). The platform figures come from each platform's own reporting, at the attribution setting you normally use. The pre-filled rows are made up.
Platform by platform
| Platform | Spend | Claims | Claimed ROAS | Share of spend | Share of claims |
|---|
If only part of your revenue came from paid
| Share of revenue from paid | Real ROAS | Profit per $100 of spend |
|---|
Real ROAS is your revenue times that share, divided by spend. The share is for you to argue about. Only a test can settle it, and Which Test Can I Run shows which tests you can do.
How it works
The checks
If the platforms together claim more than you booked, some sales were counted by more than one of them. That part is proof and needs no model. Claiming less than you booked proves nothing, because the overlap can sit inside the total.
MER treats all of your revenue as if the ads caused it, so it is the most the ads could possibly have earned. If even that ceiling is under break-even, the ads cannot pay.
The verdict
ads cannot pay: blended MER is under break-even MER. double counted: claims exceed your revenue. claims fit: neither of those, so the numbers are possible. That is not proof they are right.
Read this before you quote the numbers
MER moves with your organic mix. A brand with strong organic sales shows a high MER at any ad efficiency. Compare MER with your own earlier periods, not with another brand's.
Check the attribution window. Platforms count view-through and click-through over different windows. The claims should come from the same setting each time.
Revenue and claims must cover the same period and the same products. A mismatch makes the ratio meaningless.
This does not measure incrementality. It shows what is impossible and what is the ceiling. A holdout test shows what is real.