Two different assumptions
Redemption rate is the share of promo orders that use the coupon. Incremental orders are the orders that would not exist without the promotion. They are not the same number.
If 30% of orders redeem a 10% coupon, the discount is paid on 30% of all promo orders, including baseline orders that would have bought anyway. Those baseline redemptions are a subsidy, not growth.
The formula
Discount cost per promo order (K) = redemption × AOV × discount. Discount cost = (baseline + incremental) × K. Incremental contribution = incremental orders × AOV × contribution margin (C).
Net impact = incremental × C − discount cost − fixed cost. Break-even incremental orders solve net impact = 0: (baseline × K + fixed) ÷ (C − K). If C ≤ K, more incremental orders cannot fix the loss.
Worked example
Baseline 1,000 orders, AOV $60, 40% margin, 10% coupon, 30% redemption, 50 incremental orders, no fixed cost. K is $1.80 and C is $24.
Discount cost = 1,050 × $1.80 = $1,890. Incremental contribution = 50 × $24 = $1,200. Net impact = −$690. Break-even incremental orders = (1,000 × $1.80) ÷ ($24 − $1.80) = 81.08, rounded to 82.
Why the break-even is not 79
A shortcut divides the current discount cost by contribution per order: 1,890 ÷ 24 = 78.75, rounded to 79. That treats the discount cost as fixed, but every extra incremental order also redeems the coupon at the same rate.
The exact break-even accounts for that: 82 orders. The difference is small here, but it grows with high redemption and high discount rates. The calculator uses the exact version.
What can go wrong
A 100% redemption rate means every promo order gets the discount. If the discount per order exceeds contribution per order, no amount of incremental volume breaks even.
Fixed coupon costs, affiliate payouts and design fees also need to be covered. And if the “incremental” orders were already coming from loyal customers, the promotion is mostly cannibalization.
How to test it
Use a holdout or geo test: compare a coupon-exposed group with a similar group that does not receive it. Track redemption and incremental orders separately, then rerun the model with observed values.
A coupon code that everyone can find is not a controlled test. Use unique codes, a defined audience and a clean comparison window before treating the planning result as evidence.