Insight · MeasurementTOFU
Your ROAS is lying to you
The short answer
If you add Meta's reported revenue to Google's reported revenue and the total exceeds what your store actually made, nothing is broken. Both platforms are working exactly as designed — each claims every conversion it plausibly influenced, and a customer who saw an Instagram ad on Monday and clicked a Google ad on Thursday gets counted twice.
The fix isn't better attribution software. It's changing which number you treat as the scorecard.
How the double-count happens
Meta uses a click-through and view-through window and attributes based on its own modelling. Google Ads uses data-driven attribution across its properties. Neither can see the other. A single purchase touched by both appears in both reports at full value.
Add a retargeting campaign and it gets worse: retargeting claims sales that prospecting created. Cut the prospecting, and mysteriously the retargeting ROAS collapses too.
What to read instead
MER — total revenue divided by total marketing spend. It can't be double-counted because there's only one revenue number and one spend number. Track it weekly alongside new-customer CAC, and you'll see what your marketing is actually doing to the business.
Platform ROAS still has a job: allocating budget within a channel. Use it for that and nothing else.
The test that settles arguments
When spend justifies it, run a geo holdout: pause a channel in two or three comparable markets for two to four weeks while keeping it live elsewhere. The difference in total revenue between the markets is the channel's real contribution. It's blunt, it costs a little revenue, and it ends the debate in a way no attribution model can.
What we'd do this week
- Put MER on the weekly report, above every platform number
- Reconcile platform-reported conversions against your store or CRM
- Separate new-customer CAC from blended CAC — returning customers flatter everything
- Stop reporting prospecting and retargeting as one blended campaign ROAS
- Schedule one holdout test for next quarter
A simple double-counting example
| Source | Purchases claimed |
|---|---|
| Meta Ads Manager | 120 |
| Google Ads | 90 |
| GA4 (all channels) | 150 |
| Shopify actual orders | 160 |
Meta and Google together claim 210 purchases, but only 160 orders exist — and some came from organic, email or direct. Blended MER uses the 160 orders and total spend, so it can't be inflated by double counting.
What to do next
- Compare platform-reported revenue with back-end revenue
- Calculate blended MER weekly
- Separate brand from non-brand results
- Plan one incrementality test
Frequently asked questions
Which ROAS number should we trust?
Use platform ROAS to compare campaigns within a platform; use MER and delivered revenue for business decisions.
Why does Meta claim conversions Google also claims?
Each platform uses its own attribution window and counts conversions it influenced, including views. They don't deduplicate against each other.
How do we run a holdout test?
Pause a channel or campaign in a set of comparable regions or for a random audience share, then compare sales with the exposed group.
How do we stop double-counting conversions?
Use back-end data and blended MER as the truth, and run incrementality tests.
Keep reading
Services behind this
Terms used on this page
Talk to a strategist
Thirty minutes. Your numbers. A straight answer.
Book a strategy call with the people who'd actually run your account. We'll tell you what we'd do — or that you don't need us yet.