Why “ROAS” can lie
ROAS is simple: ROAS = revenue / ad spend. But ROAS ignores everything that makes a business real: cost of goods, shipping, platform fees, refunds and returns. That’s why you can see “2.5x ROAS” and still end the month down.
The break-even ROAS formula
Break-even ROAS answers: what ROAS do I need to avoid losing money on ads, given my margins and costs?
At a high level, you need contribution profit (from revenue) to cover ad spend:
- Contribution profit = net revenue × (gross margin − fee% − other%) − fixed variable costs
- Break-even happens when contribution profit = ad spend
If you ignore fixed per-order costs for simplicity, the quick version is:
- Break-even ROAS ≈ 1 / (gross margin − fees − other variable %)
If refunds matter, apply them to realized revenue:
- Break-even ROAS (with refunds) ≈ [1 / (gross margin − fees − other%)] / (1 − refund rate)
Example (realistic e‑commerce)
- Gross margin: 55%
- Fees: 3.5%
- Other variable %: 2%
- Refund rate: 6%
Contribution percent ≈ 0.55 − 0.035 − 0.02 = 0.495. Refund adjustment: 1/(1−0.06)=1.064.
Break-even ROAS ≈ (1 / 0.495) × 1.064 ≈ 2.15x.
So a 2.0x ROAS actually loses money here. That’s why chasing arbitrary ROAS targets can backfire.
Common mistakes (and fixes)
- Using gross revenue with VAT/tax included. Use consistent revenue definition (net if possible).
- Ignoring time lag on returns. Returns often arrive weeks later; use a rolling refund rate.
- Comparing channels with different attribution windows. Align lookback windows before declaring winners.
- Not segmenting by product. A high-margin SKU can subsidize a low-margin one — separate them.
How to operationalize break-even ROAS
Once you know your break-even ROAS, you can create a simple decision framework:
- Below break-even: optimize or pause (unless it’s intentional for new customer acquisition).
- Near break-even: acceptable for growth if LTV is strong and payback is reasonable.
- Above break-even: scale while monitoring refund rate and marginal costs.
Want a fast answer? Use the ROAS + profit estimator.