Platform ROAS is the default metric for DTC paid social. Every ads manager dashboard leads with it. Every agency report foregrounds it. Every founder pitch deck leans on it. And for most DTC brands at $2-10M revenue, it is the wrong north star. Optimizing for ROAS pulls operators toward decisions that look good on the dashboard and worse on the bank statement.
This is the contrarian argument. The alternative is not "stop measuring." The alternative is measuring against the P&L instead of against the platform.
Meta claims 4.2x. The bank shows 2.0x. The gap is what platform ROAS hides from DTC operators.
What ROAS actually claims
Platform ROAS is revenue attributed to a platform divided by the spend on that platform. Meta says 4x ROAS, it means Meta is claiming $4 of revenue per $1 you spent on Meta. TikTok says 2x, it means TikTok is claiming $2 per $1 spent. Google Ads says 3.8x on PMax, it is claiming $3.80 per $1 spent there.
All three platforms are making these claims based on their own attribution models. Meta attributes conversions to itself using 7-day click and 1-day view by default. Google attributes based on its own model. TikTok does its thing. When you sum up the platform-attributed revenue and divide by total spend, you almost always get a number much larger than your actual revenue divided by actual spend.
That divergence is not a bug. It is the design. Each platform is allowed to claim credit for conversions that would have happened anyway, for conversions that another platform could also claim, and for conversions that attribution windows extend over. Across the engagements I have audited, the sum of platform ROAS claims typically exceeds the brand's actual revenue by 30 to 60 percent, and the gap gets wider the more channels the brand runs.
The decision pattern ROAS drives
An operator looking at Meta 4x ROAS and TikTok 2x ROAS will reach an obvious conclusion: shift budget from TikTok to Meta. Meta's ROAS is twice as high. That shift often hurts the brand's total revenue. The reason it hurts is that the Meta-attributed conversions overlap heavily with Shopify organic sessions and branded search sessions, while the TikTok-attributed conversions tend to be incremental.
Cut TikTok. Meta-claimed conversions do not drop because Meta was partly claiming credit for sales that arrived through organic, email, or search anyway. Total revenue drops because the incremental contribution TikTok was making disappears. The operator sees Meta ROAS rise (since the overlap claims now have less competition) and concludes the move was correct. The bank account disagrees.
Why the industry keeps using ROAS
Three reasons. None of them are "because ROAS is the right metric."
Reason one: vendor-aligned incentives. Platforms surface ROAS because it makes the platform look like it is working. Agencies report ROAS because it is easier to show up with a 4x Meta screenshot than a nuanced blended MER conversation. Every stakeholder in the paid social economy except the brand benefits from ROAS being the default.
Reason two: cognitive ease. ROAS is one number per platform. MER is one number for the whole business. When you want to decide between platforms, per-platform numbers feel like the right tool. They are not. They are the right-looking tool.
Reason three: legacy from direct response. In classic direct response, platform ROAS was closer to honest because attribution windows were tighter and the paid channels were less overlapping with other channels. That era ended 5+ years ago. ROAS did not get updated.
The alternative: MER against the P&L
Marketing Efficiency Ratio, or MER, is total revenue divided by total paid marketing spend. It is a whole-business metric. It does not know or care which platform claimed credit. The number goes up when marketing spend is producing more revenue. The number goes down when it is not.
MER is not perfect. It does not tell you which platform is pulling weight inside the portfolio. It can hide underperformance on one channel behind overperformance on another. But MER ties directly to the P&L in a way platform ROAS never does, and the P&L is the actual business metric operators care about. MER over platform ROAS covers the blended math in detail.
Platform ROAS still has a role: diagnostic. If Meta's claimed ROAS collapses 50 percent one month, something changed on Meta. Investigate. Use the platform number to locate the problem, not to make strategic decisions.
Incrementality as the gold standard
The honest way to measure channel contribution is incrementality testing. Pause a channel for two weeks. Measure what happens to total revenue. The lift (or drop) relative to the spend is the real value of that channel. Run this once per quarter per major channel and you will have a much more accurate picture than platform ROAS will ever give you.
Most brands at $2-10M revenue are scared to pause channels because it feels risky. It is risky. It is also the only way to know what is actually working. The brands I have seen run incrementality testing routinely have a cleaner P&L and more confident media buying than the brands that have never paused a channel.
The post-iOS attribution reality
After iOS 14.5 and iOS 17, platform ROAS got worse, not better. The measurement gaps the platforms cannot see (ITP-blocked sessions, cross-device journeys, offline purchases) grew. The attribution models the platforms use to estimate these gaps became more speculative. The gap between reported platform ROAS and actual blended MER widened.
Operators who lead on platform ROAS are making decisions against estimates of estimates. Operators who lead on MER are making decisions against the bank.
“The platforms are not lying. They are claiming credit within their own model. The problem is not the claim. The problem is the operator treating the claim as the P&L.
”
The hard conversation with agencies
If your agency leads every monthly report with Meta ROAS, you are getting reported against the wrong metric. Ask them to lead with blended MER and have platform ROAS as a diagnostic section below. Most competent agencies will handle this fine. Agencies that resist usually have something to hide, which is often that the platform ROAS claim has carried their reports for a year and the blended MER tells a less flattering story.
Why most agencies overspend on top of funnel covers the agency side of this same problem from another angle.
Is ROAS useless? Should I ignore it completely?
No. ROAS is useful as a platform-level diagnostic. If Meta-reported ROAS drops 40 percent this week, something changed on Meta, investigate. The argument is against using ROAS as the north star, not against measuring it.
What MER number should a healthy DTC brand hit?
Depends on gross margin. A brand at 60 percent gross margin can be profitable at MER 2.0x. A brand at 30 percent gross margin needs MER above 3.3x to break even on contribution margin. Calculate from your specific margin structure.
How often should I run incrementality tests?
Once per quarter per major channel. More often if a channel is changing significantly. Less often if everything is stable. Any brand that has never run an incrementality test is making media buying decisions blind.
Related reading
This contrarian piece sits inside the paid social for DTC operators hub. The companion piece on the blended math is MER over platform ROAS. The agency angle is covered in why most agencies overspend on top of funnel. For the measurement plumbing that makes MER reliable, the DTC Stack Audit covers the full stack.