How DTC brands should measure blended ROAS
Most DTC brands calculate blended ROAS wrong, and the ones who calculate it right often use it wrong. It's not a marketing metric. It's not a Meta metric. It's the only number that tells you if the business actually makes money on the dollars you spend to get customers. Get the inputs wrong and every downstream decision — budget allocation, channel mix, CFO conversations — is built on sand.
What blended roas actually measures
Blended ROAS is total revenue divided by total marketing spend across every paid channel, over the same time period. Not platform-reported revenue. Not last-touch attributed revenue. Total revenue from your order data, divided by everything you spent to acquire and retain — Meta, Google, TikTok, affiliate, influencer, SMS, email platform fees if you're counting them.
The point of blended ROAS is that it doesn't care about attribution models. It doesn't ask which channel "deserves credit." It answers one question: for every dollar spent on marketing, how many dollars came back. That's the question your P&L cares about, and it's the question platform-reported ROAS actively obscures because every platform inflates its own contribution.
If Meta says 4x, Google says 3x, and TikTok says 2.5x, but your blended ROAS is 1.8x, you don't have a measurement problem — you have an overlap problem. Those platforms are all taking credit for the same conversions.
The time window problem nobody fixes
Here's where most brands get blended ROAS wrong even when they're calculating it "correctly." They match spend and revenue to the same calendar period — spend in March divided by revenue in March. That's fine for a subscription box with instant conversion. It's wrong for anything with a consideration cycle, which is most DTC.
If someone clicks an ad on March 28th and buys on April 3rd, that revenue belongs to March's spend, not April's. Calendar-month blended ROAS smears this across periods and makes your numbers noisy in ways that hide real trends. A brand running a big push in the back half of a month will look artificially weak that month and artificially strong the next, even if efficiency never changed.
The fix isn't complicated, it just requires actual data infrastructure: build a rolling window model. Attribute revenue to spend based on a realistic conversion lag for your business — 3 days, 7 days, 14 days, whatever your funnel supports — using order timestamps and first-touch or click data, not just the month the invoice landed in. This is a warehouse job, not a spreadsheet job, because you need order-level data joined to session or click data with real timestamps, not aggregated platform exports.
New customer roas vs blended roas
Blended ROAS answers "are we profitable overall." It does not answer "should we keep spending on this channel." Those are different questions and conflating them is how brands either overspend into diminishing returns or panic and cut channels that are actually working.
The reason: blended ROAS includes returning customer revenue, and returning customer revenue is largely a function of past acquisition spend and retention work, not this month's ad budget. A brand with a strong repeat purchase rate can show a great blended ROAS while new customer acquisition is quietly bleeding money, because loyal customers are propping up the average.
You need both numbers, split cleanly:
- Blended ROAS — total revenue over total spend, the health check for the whole business
- New customer ROAS (nCROAS) — revenue from first-time buyers only, over acquisition spend, the health check for whether you can keep growing
If blended ROAS is healthy but new customer ROAS is falling, you're running on fumes from a customer base you're not replenishing. That's a warning sign that looks fine on the surface and gets ignored until growth stalls.
Building the pipeline that makes this real
None of this works off platform dashboards. You need a data pipeline that pulls three things into one place with consistent grain: order-level revenue from your ecommerce platform, spend by channel and campaign from every ad platform, and a customer flag for new versus returning tied to actual purchase history, not a platform's guess.
Practically, that means:
- A warehouse (BigQuery, Snowflake, or even a well-structured Postgres instance) as the source of truth
- Daily or near-real-time syncs from ad platforms via their APIs, not manual CSV exports that go stale
- Order data joined by customer ID with a clean first-purchase-date flag so new/returning is calculated from your own records, not inferred
- A rolling attribution window baked into the model, not bolted on in a spreadsheet after the fact
Once this exists, blended ROAS and new customer ROAS become queries, not monthly fire drills involving three people and four exported spreadsheets that don't reconcile. That reliability matters more than people admit — a number you have to rebuild by hand every month is a number that quietly gets fudged, skipped, or trusted less than it should be.
What to actually do with the number
Set a blended ROAS floor based on your margin structure, not a competitor's benchmark or an agency's target. If your gross margin is 60%, you can afford a lower blended ROAS than a brand running at 35%. Calculate your breakeven ROAS from contribution margin first, then treat everything above it as the range you're optimizing within.
Track blended ROAS weekly on a rolling basis, not monthly on a calendar basis. Track new customer ROAS alongside it, always. And when the two numbers diverge — blended holding steady while new customer ROAS drops — treat that as the leading indicator it is, not noise to wait out.
Blended ROAS isn't a vanity metric or a marketing scoreboard. It's a solvency check. Build the pipeline to calculate it honestly, pair it with new customer ROAS so you're not fooling yourself, and you'll make budget decisions based on what the business actually does with a dollar — not what a platform claims it did.