Meta Says 4x ROAS. Your P&L Says You Lost Money. Your P&L Is Right.

It is Monday morning. A Head of Growth sits down to her QBR with two numbers that both claim to be true. Meta's dashboard says the account did 4x ROAS last month. The CFO's P&L, which arrived Friday, says the brand lost money on the same spend.

Both numbers are real. Both are pulled from actual systems. And she is about to spend forty-five minutes explaining why they disagree.

The gap is not a tracking bug. It is not a missing UTM or a broken pixel. One of those numbers is graded by a party with skin in the game, and the other is graded by the bank. Those are not the same exercise.

Every Platform Is the Scorekeeper and the Player

A platform's ROAS is not a measurement. It is a claim the platform makes about its own performance.

Meta does not report your ROAS from a neutral seat. It builds the attribution model, sets the lookback window, decides whether a view counts, and decides what qualifies as a conversion. Then it runs the numbers and hands you the result. No external auditor is involved.

The mechanics are plain. Each platform credits conversions inside a window it defines: typically seven days for a click, up to one day for a view by default, adjustable from the ad account. Widen the window and the same conversions get counted more. Add view-through credit and impressions that may have contributed nothing start generating revenue entries. After 2021's ATT changes cut off much of the deterministic signal, more of what a platform reports became modeled rather than observed, which loosened the number further.

None of this is fraud. These are legitimate attribution choices, and different choices suit different businesses. The problem is not that the platform made choices. The problem is that nobody told you the platform was choosing.

That is what grading your own homework means. The number goes up when the platform adjusts a lookback, adds view-through credit, or models a conversion more generously. The revenue in your bank account does not move with any of it. Your CFO reads the bank.

Three Platforms, One $80 Order

Here is the double-counting problem in its simplest form.

A customer sees a Meta ad on Thursday. Friday she sees a TikTok ad for the same product. Saturday she types the brand name into Google and clicks the branded search. Sunday she buys. One order, $80.

Meta claims the sale. TikTok claims the sale. Google claims the sale. Add the three dashboards and the brand appears to have sold the product three times. The bank recorded one transaction.

This is not a corner case. It is what cross-channel attribution looks like when every platform runs its own model, counts conversions inside its own window, and never reconciles against what the others are counting. There is no shared ledger. The conversion happens once; the credit gets taken three times.

In-platform ROAS figures do not net to a total. Add up the revenue your platforms claim they drove and you will routinely land above your actual revenue, sometimes well above it. Blended MER does net to a total. Marketing efficiency ratio is total revenue divided by total marketing spend across every channel. It carries no attribution model and makes no claim about which channel caused which sale. It is the arithmetic your CFO does when she reconciles the bank, and it cannot be gamed because it has no variables to tune.

How to Reconcile When the Platform Disagrees With the Bank

The reconciliation is simpler than most brands expect, and the finding is almost always uncomfortable.

Start with three columns: the platform-reported revenue from each channel, your GA4 purchase revenue for the same period, and your blended MER. Line them up.

The gap between platform-reported revenue and GA4 is where the over-attribution lives. GA4 is not administered by any ad platform and is not invested in any attribution outcome. It is a source of truth the platforms do not control, which is exactly why measuring against it tells you something.

As Curtis Howland, a DTC growth consultant advising brands at $30M+ in annual ad spend, puts it: "Every platform is grading its own homework. And every platform gives itself an A+. When Meta says ROAS is 4x and your P&L says you lost money, your P&L is right."

For most brands running this comparison the first time, the platform-reported total exceeds GA4 by twenty to forty percent. The gap was always there. Nobody had put the two columns side by side.

The MER column is the final check. Because it starts from one revenue number, it is immune to double-counting. When MER is healthy, the business is probably working even if individual platform figures run hot. When MER is falling while every dashboard looks good, you are watching the gap grow in real time. Cody Plofker, who ran growth at Jones Road Beauty, said it flat: "MER and nCAC are the only metrics that can't lie to you." They cannot lie because nobody selling you ad inventory administers them.

In-Platform ROAS Still Matters. Just Not for Grading.

The honest objection deserves a straight answer.

In-platform ROAS is useful for steering inside the platform. The algorithm optimizes against the signal it can see, so when you are deciding which ad set to scale or which creative is pulling weight, the in-platform number is the signal the machine is learning from. Using it there is reasonable. That is what it is for.

The error is promoting an internal steering signal to the scoreboard you judge the whole business on. The number is right for driving. It is not right for grading.

MER, for its part, is coarse. It will not tell you whether Meta is beating TikTok, and it was never meant to. The point is not to replace channel diagnostics with MER. The point is to reconcile every channel's self-reported claim against the one number that reflects what the business actually did, then investigate the gap. Blended is the truth check, not the whole analysis. Use in-platform ROAS to steer inside each channel; use MER and GA4 to grade the total.

What This Means for a $30M+ Brand

A brand at $30M+ with a serious paid budget should be graded against a source of truth the platform does not administer. In practice that means a standing reconciliation, platform revenue versus GA4 versus MER, in the same document, at the same QBR where the dashboards come out. The gap becomes visible, and decisions made against a visible gap beat decisions made against a hidden one.

It matters most when you are judging whether your growth partner is working. Anyone paid a percentage of your ad spend has no reason to surface the gap between platform-reported ROAS and MER, because one of those numbers is tied to their fee and the other is not. A party with no stake in the spend number is the one positioned to show you the reconciliation and let you draw your own conclusions.

Sutton was built to encode that discipline. $150M in DTC sales driving 6 exits across our founding team is the record behind it, and the grading standard that came out of that experience is the one we still run: your own GA4 and MER, the numbers your CFO believes, not the platform's self-reported ROAS, with the gap shown to you whenever the two disagree.

The QBR, Revisited

The Head of Growth who opened with two contradictory numbers is not in an unusual spot. Every serious brand with cross-channel spend is in the same place. The dashboard and the P&L measure different things, and only one of them started from the bank.

The rule she took away, once she ran the reconciliation: when the platform says 4x and the P&L says loss, trust the P&L. Use the platform's number to drive the machine. Grade the machine on the number the platform does not control. That is not a novel insight. It is what your CFO has been doing all along.