Why it matters
ROAS is the default efficiency metric in performance marketing. Teams set targets, pace budgets, and compare channels on it daily. But ROAS is not one number. Meta, Google, and TikTok each attribute revenue inside their own windows and rules. Finance may use cash collected in the bank. Analytics may use cohort LTV at D90. Same spend, different numerators, different stories.
The gap matters most when customer value is delayed. A campaign can show strong ROAS on first-order purchase value while repeat buyers and refunds erode profit weeks later. Conversely, value-based bidding on predicted lifetime value (pLTV) may look conservative in platform ROAS early on while acquiring higher-quality users who mature later.
Performance teams need to name which ROAS they mean in every report. Without that discipline, platform wins get scaled while blended or incremental readouts stay flat.
Return on ad spend (ROAS)
ROAS is often the success metric for value-based bidding pilots, but the numerator must match what you want the platform to optimize:
- First-party data and revenue history in your data warehouse define gross, net, and margin views of customer value.
- Churney models user-level pLTV and sends calibrated values directly to ad networks via Meta CAPI and the Google Ads Conversion API.
- Campaigns run value optimization, target ROAS (tROAS), or equivalent value goals using those signals.
- During the pilot, track platform ROAS for delivery health and blended ROAS for channel mix context.
- Judge scale decisions on incremental ROAS vs business as usual (BAU) at agreed cohort maturity, not platform totals alone.
The data warehouse feeds modeling and readout. Churney activates the value signal. Your experiment design owns whether ROAS improvement is causal lift or attribution reshuffling.
Generic form:
ROAS = Attributed revenue / Ad spendVariants to label explicitly:
| Label | Numerator | Denominator | Typical use |
|-------|-----------|-------------|-------------|
| Platform ROAS | Conversion value in platform reporting | Platform-reported spend | Daily campaign pacing |
| Blended ROAS | Revenue from blended or internal attribution | Total or allocated ad spend | Cross-channel efficiency |
| Incremental ROAS | Incremental revenue from test vs control | Incremental spend | Pilot and causal readout |
Interpretation guardrails:
State revenue basis (gross, net of returns, margin).
State attribution window and source.
Pair ROAS with volume; efficiency can rise while scale falls.
Category variants
| Model | How ROAS shows up |
|---|---|
| Ecommerce / DTC | Platform ROAS on Purchase events; blended ROAS across Meta and Google; returns and discounts often absent from platform numerators until modeled. |
| Subscription app | Trial-start or subscribe ROAS on short-window value; true economics need renewal and churn in maturity readout. |
| SaaS / PLG | Signup or PQL conversion value as numerator; expansion revenue may land outside default attribution windows. |
Common mistakes
- Treating all ROAS as comparable. Platform, blended, and incremental ROAS answer different questions; mixing them in one slide deck confuses stakeholders.
- Using gross purchase value when net margin is the goal. Refunds, discounts, and COGS can flip a "winning" ROAS into a loss at maturity.
- Optimizing platform ROAS without incrementality testing. Attribution numerators can rise when the algorithm finds cheaper clicks on users who would have converted anyway.
- Short readout windows on LTV businesses. First-order ROAS peaks before repeat and refund patterns stabilize.
- Flat value on every conversion. ROAS goals tied to uniform event values behave like volume optimization, not value-based bidding.
- Scaling on platform ROAS alone after a pLTV change. Pair dashboard ROAS with holdout or structured experiment readout before budget shifts.
Advertiser lens
| Role | What they ask | What good looks like |
|---|---|---|
| Head of Performance / UA | What ROAS target should we set? | Explicit definition (platform vs blended), revenue basis (gross vs net), and attribution window documented per channel. |
| VP Growth / CMO | Are we efficient enough to scale? | ROAS trend plus volume, customer quality, and incremental readout at maturity. |
| Marketing Analytics / Data Science | Which ROAS is "true"? | Separate reporting for platform, blended, and incremental; agreed primary metric for pilots. |
| Data Engineering | Can we reconcile platform and internal ROAS? | Join keys from ad platforms to order and subscription systems; clear revenue definition in the data warehouse. |
| Finance / Procurement | Does ROAS map to contribution margin? | Numerator aligned to margin or net revenue where possible; pilot success criteria pre-registered. |
FAQ
What is return on ad spend (ROAS)?
ROAS is attributed revenue from advertising divided by ad spend. It expresses how much revenue (per your chosen attribution and revenue definition) you earn for each dollar spent on ads.
How do you calculate ROAS?
ROAS equals attributed revenue divided by ad spend. Example: $40,000 attributed revenue and $10,000 spend yields ROAS of 4.0, often written as 4x or 400% depending on internal convention.
What is a good ROAS?
There is no universal benchmark. A "good" ROAS depends on margin, payback period, incrementality, and whether the numerator is first order or lifetime value. Finance should set targets from unit economics, not industry averages alone.
How is ROAS different from ROI?
ROAS uses revenue in the numerator. ROI typically uses profit or margin after costs. A high ROAS can still be unprofitable if margin is thin or returns are high.
Why does platform ROAS differ from blended ROAS?
Platforms credit revenue inside their own attribution models and windows. Blended ROAS aggregates across channels, tools, or internal attribution rules. Overlap, view-through credit, and cross-device gaps create systematic differences.
Can value-based bidding improve ROAS?
It can improve the efficiency of spend toward higher-value users when value signals are accurate, fresh, and matched. Validate with incremental ROAS and cohort readout, not platform dashboard movement alone.
Should pLTV pilots use platform or incremental ROAS?
Use platform ROAS for delivery monitoring during the test. Use incremental ROAS at cohort maturity for scale and budget decisions.
Not the same as
| Term | Difference |
|---|---|
| Platform ROAS | One network's attributed numerator; subset of ROAS definitions. |
| Blended ROAS | Cross-channel or internal-model ROAS; not platform-specific. |
| Incremental ROAS | Causal lift numerator; requires experiment design. |
| Cost per acquisition (CPA) | Spend per conversion; inverse framing of efficiency, not revenue ratio. |
| Customer acquisition cost (CAC) | Often fully loaded acquisition cost per customer; may include non-ad costs. |
