Referral Rewards, Ranked by Cost Per Real Customer

Referral Rewards, Ranked by Cost Per Real Customer

Referral programs look cheap: you pay only when someone converts. But the sticker price hides three costs — fraud, non-incrementality, and the giver's reward economics — that routinely flip which reward you should offer.

The three reward types, honestly priced

1. Cash (or true cash-equivalent). Highest perceived value per dollar, so it converts sharers well. But cash has no breakage, no brand lock-in, and attracts gaming (self-referrals, manufactured "friends"). The unit economics are exactly what they look like — plus fraud on top.

2. Store credit. The middle ground, and usually the winner on true cost. Credit keeps spend in-house, carries breakage (a meaningful share is never fully redeemed), and self-selects for customers who actually want your products. Downside: fulfillment still costs COGS, and credit balances are a liability to book.

3. Percentage discount on next order. Cheapest-looking, often most expensive in reality. "$20 off your next order" costs face value off margin, trains the referred customer to expect discounts from day one, and — worst — travels: referral codes leak to coupon sites like any other code, acquiring bargain hunters at your expense.

The math that matters

True cost per real referred customer:

(giver reward + receiver cost + overhead) ÷ (referred orders that are both legitimate and incremental)

The denominator does the damage. Subtract self-referrals, duplicate-household gaming, and buyers who would have purchased anyway (referred by a friend who was already recommending you). Referral programs commonly see meaningful leakage on that denominator — enough to turn a "$40 per acquired customer" headline into $55–60 true cost.

Then compare to paid CAC on equal windows (60–90 day, refund-adjusted) — and compare referred-customer LTV to paid-acquired LTV. Referred customers often retain better, which can rescue a program that looks pricey on CAC alone.

The controls that make it work

Before scaling, code these in:

• New-customer only, with payment/device checks.
• Per-giver monthly caps.
• Minimum AOV for the reward to trigger.
• No stacking with other codes.
• Giver payout after the friend's return window closes.

Every control tightens the denominator — which is where referral programs actually lose money.

The ranking (typical case)

1. Store credit — retained in-house, breakage-adjusted, locks in a return visit.
2. Cash — honest and simple, but leaks out of the ecosystem with the highest fraud pull.
3. Percentage discount — cheapest to offer, most expensive in training and margin leakage.

Your mileage varies with AOV and margin — so run the true-cost math on your own numbers, against a paid-CAC baseline, before choosing.

The takeaway

Referral isn't "free acquisition." It's acquisition with three hidden taxes: fraud, non-incrementality, and whatever the reward teaches the customer.

Price the reward on true cost per legitimate incremental customer — and make it store credit wherever you can, so the reward is the next sale.