Your ROAS is lying to you: an attribution field guide
Cristian Bragau, Founder & Lead Auditor · Reviewed · verified against the base of 74 auditsNot as reported. In our audits, the revenue the ad platforms claim overstates real blended return by 20 to 30% on average, because Meta, Google and GA4 each count the same sale. Attribution inflation appeared in 54% of the 74 audits. The fix is one line: total revenue divided by total marketing spend.
Why every platform claims the same sale
In our audits, reported ROAS overstates real blended return by 20 to 30% on average. Not because anyone is cheating, but because Meta, Google and GA4 each claim the same sale. One DTC client’s dashboards summed to 122% of actual revenue.
Each platform measures the customers it touched. A shopper who clicks a Meta ad on Monday, searches the brand on Google on Thursday and buys from an email on Friday is one sale, three touches and, in three separate dashboards, three conversions. None of the platforms is wrong by its own rules. The rules were simply never built to add up.
Attribution windows overlap
Meta’s default window counts a purchase within seven days of a click or one day of a view. Google Ads counts conversions up to 30 days after a click by default. GA4 applies its own model on top of both. The windows run at the same time over the same customers, so the longer and wider they are, the more of the same revenue each one can claim.
Branded search takes the credit
Branded search and retargeting sit closest to the purchase, so they collect the last click on sales other channels created, and on sales nobody created at all. That is why they report the highest ROAS in most accounts, and why non incremental spend was the most common finding in the 74 audits, at 68%.
Three checks you can run this week
Add up the revenue each platform claims for one month and divide by actual revenue. Above 110% you’re double counting (most audits land at 120 to 160%) and overfunding whichever platform claims most aggressively.
Compare branded search and retargeting spend against what organic and CRM would capture anyway. In our marketplace audits, up to 29% of paid search bought clicks the brand already ranked #1 for.
Rescore campaigns on contribution margin (profit after all variable costs) per SKU or cohort, not blended ROAS. In 8 of 10 ecommerce audits, 20 to 30% of spend was buying revenue below breakeven after shipping and returns.
What the 74 audits found behind reported ROAS
| Measure | Across the 74 audits | What it means for ROAS |
|---|---|---|
| Non incremental paid spend | 68% of audits | The platforms report sales you would have made anyway as wins |
| Attribution inflation | 54% of audits | The same sale is counted by more than one platform |
| Wasted paid spend | 22% of spend | Media that produced no incremental margin, whatever its ROAS |
| Effective spend | 62% of spend | The share of the budget that did the job it was paid for |
| Median leak share | 18.7% of spend | What the typical budget was losing while its dashboards read green |
Use MER to steer the budget
The fix for double counting is a number nobody can double count. MER divides total revenue by total marketing spend: one line, no attribution model, no platform deciding who deserves the credit. It is the same calculation as the real blended return in the chart above.
MER will not tell you which campaign to cut. It tells you whether the whole budget is working, and it moves when a cut was real. Use the platform reports to choose what to test, and MER to judge the result. When a platform reports a record month and MER stays flat, the record belongs to sales another channel already had.
Then score campaigns on margin
Revenue is the wrong scoreboard for a campaign. A sale whose shipping, returns and payment fees cost more than its margin loses money at any ROAS, a margin leak no attribution report will show. Rescore your top campaigns on contribution margin, then compare each one with its own breakeven. The arithmetic: at a 40% contribution margin, breakeven is a 2.5× return; at 25%, it is 4.0×. In Scaling prospecting spend was losing money on every sale, two prospecting campaigns looked profitable on ROAS and were buying revenue below breakeven.
Fixing attribution rarely recovers much money directly (a median of $3k). But it changes where every other dollar goes: it’s the finding that makes the other findings visible. All three checks run on exports you already have; the audit runs them (and 34 more) against your verified revenue data.
Questions about ROAS and attribution
Why does my ROAS look higher than my actual results?
Because each platform counts the sales it touched, and most sales are touched by several. Meta, Google and GA4 each claim the same purchase, so the revenue they report together adds up to more than the revenue you banked. Attribution inflation appeared in 54% of the 74 audits. Attribution inflation, defined →
What is the difference between ROAS and MER?
ROAS divides the revenue one platform claims by that platform’s spend. MER divides your total revenue by your total marketing spend. MER cannot double count, because it never asks a platform who deserves the credit, which is why budget decisions should be judged on it. MER, defined with a calculator →
What is a good blended ROAS?
There is no universal number, because breakeven depends on your contribution margin. A business keeping 40 cents of every revenue dollar after variable costs breaks even at 2.5×; one keeping 25 cents needs 4.0×. Judge each campaign against your own breakeven, not against an industry average. Blended ROAS, defined →