Add up the ROAS that every platform reports and you'll often "earn" more revenue than your bank actually received. We see the summed-up version overstate real sales by roughly 2x on multi-channel accounts. The platforms aren't broken. They're each taking full credit for the same sale.
That is the whole reason why platform ROAS is misleading: it was built to defend the platform's budget, not to tell you whether your business made money.
What we actually looked at
This is not a single dataset. It's the pattern we see across DTC, ecom, and info-product accounts we audit - usually the ones running Meta and Google together, sometimes with email and affiliates stacked on top. The shape is always the same: the in-platform dashboards glow, the blended number sags, and nobody can square the two.
The figures below are illustrative benchmarks, not one client's results. They show the mechanism, not a case study. The mechanism is what matters, because it repeats everywhere.
Every platform counts the same sale as its own
Attribution is not neutral. Meta counts a conversion if its ad was seen or clicked inside the attribution window. Google counts the same conversion if a search or a YouTube view touched the journey. Email counts it if the buyer opened a flow. One sale, three systems, three full credits.
Stack those numbers in a spreadsheet and they sum past 100% of reality. This is attribution double counting, and it is not a glitch. It is each platform doing exactly what it was designed to do: prove its own worth to keep your budget.
The bars sum to $180 of "attributed" revenue against $100 that actually landed. Now imagine that distortion running across an entire account, every day, into the report your CEO reads.
The gap widens exactly when you scale
At small spend the overlap is small, so the lie is cheap. The problem is that the gap between platform roas vs blended roas grows with budget. More channels, more retargeting, more brand spend - more overlap for the platforms to fight over.
So the brands most exposed are the ones scaling hardest. The account looks more profitable on the dashboard precisely as it gets more dangerous in the bank.
If your reporting only shows platform ROAS, you are paying to be lied to. Politely. By an algorithm with a quota.
Blended ROAS is the number that can't double-count
There is one number the platforms cannot inflate, because they don't get to touch it. Blended ROAS is total revenue divided by total ad spend, pulled from your own books, ignoring who claimed credit.
It can't double-count because there is nothing to count twice. One revenue figure, one spend figure, one ratio. If you pulled $500k against $100k of spend, blended ROAS is 5 - and no platform gets a vote.
This is the same idea as the Marketing Efficiency Ratio, just expressed as a multiple instead of total spend. We unpack the full three-number stack in ROAS vs MER vs CAC, because platform ROAS is only one altitude of the picture.
Where this read falls apart
Blended ROAS has a real weakness, and pretending otherwise would make this post the same kind of lie. Blended can't tell you which channel is working. Kill your best-performing campaign and blended ROAS barely twitches for days, because the rest of the spend masks it.
So platform ROAS is not useless. It's a diagnostic. It tells you which lever moved at the campaign level. The error is promoting a diagnostic into the headline number that decides whether you scale. Read it as a clue, not a verdict.
That distinction also drives the question everyone asks next - what is a good ROAS - because the answer depends entirely on which ROAS you mean and what your margin can carry.
What to track on Monday instead
Put both numbers in the same report, in the same row, with the gap as its own column. The gap is the story - not either number alone.
- Blended ROAS as the headline. Total revenue over total spend, from your books.
- Platform ROAS as the diagnostic, per channel, clearly labelled as in-platform.
- The gap between them, tracked weekly. A widening gap is your early warning that you're scaling a mirage.
When the gap quietly doubles, that's the signal to slow down - and it's exactly the kind of drift that hides in a monthly report until the cash is already gone. Our always-on layer scans the stack every morning at 7am and flags the day the two numbers stop agreeing, so the call gets made by a human while the spend is still recoverable. The kind of weekly discipline we build into every account we run starts here.
If you scaled last quarter on platform ROAS alone, which number were you actually trusting - the one on the dashboard, or the one in the bank?