What platform-reported ROAS measures
Every advertising platform divides the revenue it believes it caused by the money you spent inside it. The spend half is reliable, because the platform billed you for it. The revenue half is a claim, built from that platform’s own attribution rules: which clicks it counts, whether it counts views as well as clicks, how long after the ad it will still take credit, and how much of the unobserved behaviour it fills in by modelling.
Those rules differ between platforms, and so does the date the revenue lands on. Google Ads reports a conversion against the date of the click that earned it, so yesterday’s figures keep rising for days afterwards. GA4 records it on the day the purchase happened. Meta credits within its own click and view windows. None of them is lying; they are answering different questions.
Why platform-reported ROAS overstates performance
A single sale is usually touched by more than one channel. Someone sees a Meta ad, searches your brand, clicks a Google ad, then buys. Each platform sees its own touch, counts the full order value, and reports it. Add the platforms together and the total revenue exceeds what the business actually banked, sometimes by a wide margin.
Retargeting makes it worse, because retargeting audiences are made of people who were already going to your site. A platform will happily claim those sales, and its reported return on ad spend will look excellent, while the incremental contribution — the sales that would not have happened otherwise — is far smaller.
Common mistakes with platform-reported ROAS
The first is summing channel revenue into a single number for a board report. The second is judging channels against each other on their own reports, when each is scored by a different referee; a channel that mainly runs prospecting will always look poorer than one running retargeting, regardless of which one grew the business.
The third is treating the gap between platform revenue and accounting revenue as a tracking fault to be fixed. Some of it is a real fault worth chasing, but a large part of it is double counting and attribution windows behaving exactly as documented.
How to act on it
Set one number as the truth for money: total revenue from your accounts or order system, divided by total advertising spend across every platform. That blended figure cannot be inflated by double counting, and it is the one that pays wages.
Use platform-reported figures for decisions inside that platform — which campaign, which audience, which creative to keep — where the rules at least stay consistent. Note the attribution model and window on every report you send, so the reader knows what they are looking at. When a budget decision is large enough to justify it, test by holding a channel back and watching what happens to overall revenue, rather than by arguing about whose report is right.