Why blended ROAS beats platform ROAS, every time.

Here's a thing we've watched happen, in some form, on every paid media account we've audited in the past three years: a marketing team reports stellar platform ROAS, Google says 4.5×, Meta says 3.8×, TikTok says 7.2×. The CFO does the math, asks the obvious question, and the room goes quiet. If every channel is earning $4 for every $1 spent, where's the $4 actually going? Why isn't gross margin tracking the way the dashboards say it should?

The answer, every time, is that platform ROAS is not what most marketers think it is, and the conversation with finance is breaking down because two parties are looking at fundamentally different numbers and assuming they mean the same thing.

What platform ROAS actually measures

When Google Ads tells you a campaign earned 4.5× ROAS, it means: of the conversions Google's attribution model credited to this campaign, within Google's attribution window, using Google's view of the conversion funnel, the value summed to 4.5× the spend. That's it. Each clause is doing real work, and each clause is also where the number gets generous.

• "Conversions Google's model credited." Default attribution is data-driven, which spreads credit fractionally across all touchpoints Google can see. But Google can only see what happens in Google's ecosystem (and adjacent sites with the GA4 tag). Touchpoints on Meta, on email, on direct, Google doesn't know about them. So Google takes more credit than it deserves, structurally.
• "Within Google's attribution window." Default is 30 days. Many businesses have longer consideration windows than 30 days, which means a click that genuinely contributed to a conversion 45 days later gets under-credited. But for most of paid search and PMax, this generally tilts in Google's favor too, the "free" conversions from in-market intent get attributed to the last paid click.
• "Of the conversions tracked." This is the big one. Conversions Google can't track, phone calls without call extensions, in-store visits without offline conversion uploads, email-driven conversions, attributable assists Google never sees, don't show up. So when these are happening (and they always are), Google's model has to inflate credit on the conversions it can see to balance to 100%.

Same logic, with different specifics, applies to Meta's attribution (which is even more aggressive, view-through credit on a 1-day post-view window will inflate ROAS substantially), TikTok's (loose), Amazon's (genuinely closed-loop, but only for in-Amazon transactions), and so on.

Each platform is grading its own homework, on a curve, with the answer key.

The double-counting problem

Here's the practical consequence: if you sum platform ROAS across channels, you're double-counting. Triple-counting, in many cases. The same conversion gets credited fractionally to Google, fractionally to Meta, fractionally to TikTok, and partly to email, and each platform reports its share back to you as if it were a closed system.

Take a hypothetical: $100K monthly spend split $40K Google / $30K Meta / $20K TikTok / $10K email. Platform ROAS reports 4.5× / 3.8× / 7.2× / 12×. Naively summed, that's $180K + $114K + $144K + $120K = $558K of attributed revenue. Actual tracked revenue from the GA4 e-commerce report? $290K.

The platforms collectively claim 1.92× the revenue that actually happened. That is not a small reporting error. That is the difference between a campaign you'd scale aggressively and one you'd cut.

What blended ROAS measures, instead

Blended ROAS is structurally simple: total tracked revenue ÷ total marketing spend, across all paid channels and (depending on how you frame it) all marketing-eligible spend including agency fees, content production, and tooling.

It's deliberately blunt. There's no attribution model. There's no per-channel credit. There's just: how much money came in, how much went out, what's the ratio. The CFO's view of the world.

For most B2C businesses, the meaningful blended ROAS is somewhere between 1.5× and 5×, depending on how aggressively you draw the spend boundary. Anything north of 3× across a meaningful spend base is a healthy program. Anything below 1.5× is bleeding money.

Compared to platform ROAS, which can read 6× on a campaign that's actually destroying value once you net out the platform's over-attribution, blended is real. Painfully real. Which is exactly the point.

How to actually calculate it

The principle is simple. The execution depends on what infrastructure you have. In our work, we typically build blended ROAS into a single weekly report with three columns:

01Total marketing spend

Pulled from each ad platform's billing report (not the in-platform spend dashboard, which sometimes diverges from billed amount). Normalize to a single currency, single time zone, single fiscal week. Add agency retainers, content production, tools, and any other dollars whose justification is "marketing."

02Total tracked revenue

From GA4 (for eCom), from the e-commerce platform's order log (more accurate, no sampling), or from the CRM (for B2B with longer sales cycles). For B2B, count pipeline-stage revenue at probability-weighted values, and net-revenue-retained on existing customers if your motion is land-and-expand.

03The ratio, and the trend

Compute blended ROAS for the week. Plot it as a 4-week moving average against the same period prior year. The trend line is more diagnostic than any single point.

The friction is data assembly, not math. Most teams that haven't done this before are surprised by how much manual work it takes the first month. By month three it's automated, and the question becomes "why didn't we always look at this?"

The conversation shift

Once blended ROAS exists in your reporting, the marketing-finance conversation changes character. It shifts from defending platform metrics ("but Meta says 3.8×!") to triangulating drivers ("blended dropped 0.4× this month, which channels saw spend lifts that didn't pull through?").

This is a much better conversation. It's the one finance was trying to have all along. And it positions marketing as a discipline that owns its P&L, not a cost center that performs interpretive dance with platform dashboards.

The objections (and our usual answers)

→"But platform ROAS still matters for in-platform optimization."

Yes, for the campaign manager tuning bid strategies, audience targeting, and creative rotation, platform ROAS is the right operational metric. The point isn't to abolish it. The point is to stop reporting it up. The C-suite scoreboard is blended. Channel-level operators tune to their platform numbers; the executive view is one tier higher.

→"Blended doesn't tell us where to allocate."

This is technically true and practically wrong. You allocate by running incrementality tests, geo lift, holdout cohorts, controlled-burn experiments, that show you the true incremental return per channel. Then you scale the channels that pull blended up. Platform ROAS gives you a directional signal at best for this; incrementality tests, applied to a blended reporting frame, give you the real answer.

→"Our boards have always reported platform ROAS."

So have most boards. Habit is not evidence. We've yet to meet a finance team that didn't feel relief once the conversation moved to blended. The marketers fight it harder than the CFOs do, usually because the platform numbers tell a more flattering story.

What to do this quarter

If your reporting today is platform-centric and the CFO conversations are tense, here are three concrete moves:

Blended ROAS isn't a silver bullet. It's just the metric that both the marketing team and the finance team can agree means the same thing, calculated the same way, in the same direction. In our experience that alignment is worth a lot more than the precision of any individual platform dashboard. It's the metric the work actually gets graded on.