Why blended ROAS misleads
Most ecommerce brands track ROAS at the channel level: revenue ÷ ad spend. A campaign reports 4.0× and the team scales it. Three months later the P&L is somehow worse, even though every dashboard is green. The reason: blended ROAS measures gross revenue, not the contribution that revenue earned after the cost of producing it.
True ROAS subtracts the costs ad spend incurred to generate the revenue — COGS, returns, shipping, and payment fees — before dividing by ad spend. The result is the actual dollar contribution per dollar spent. If true ROAS is below 1.0×, the channel is losing money no matter how impressive the gross ratio looks.
The formula
True ROAS = (Revenue × (1 − COGS%) × (1 − Returns%) − Revenue × Shipping% − Revenue × Fees%) ÷ Ad spend.
The order matters. COGS is incurred before returns reduce sellable inventory; returns reduce both the revenue and the COGS already paid (we approximate this as a single revenue-reduction term, which is conservative for most apparel + accessory brands).
How to use the output
- True ROAS < 1.0×: the channel loses money on the marginal sale. Cut or rework before scaling.
- 1.0× – 1.5×: break-even territory. Acceptable only if LTV recovers the gap; otherwise rework.
- 1.5× – 2.5×: healthy. Scale carefully and watch new-vs-returning mix.
- > 2.5×: strong. Scale aggressively unless the channel is volume-capped.
Common mistakes the calculator surfaces
Forgetting payment fees. A 2.9% Stripe fee on a 4× ROAS campaign reduces true ROAS by roughly 0.12. Small per-order, large across an account.
Ignoring returns. A campaign that drives impulse buys often runs 12–18% return rates. At 15% returns, a 3.5× campaign becomes 2.9× true — and that's before reverse-logistics costs.
Using last-touch revenue. The revenue input should be the revenue this channel uniquely drove, not the total touched. Multi-touch attribution closes that gap; if you don't have it, run the calculator twice — once with blended-revenue, once with platform-reported — and treat the spread as your attribution-risk band.