Referral programs pay both the existing customer and the new one, which looks like paying twice for a single sale. The arithmetic works because of what the alternative costs.
The comparison is against paid advertising
Every company knows roughly what it spends to acquire a customer through search, social and display advertising. That figure is the budget a referral has to beat.
A combined referral payout that lands under that number is cheaper acquisition, even though the money is visible to the customer instead of disappearing into an ad auction.
It is also more certain. Advertising buys impressions with an uncertain conversion rate, while a referral bonus is paid only when a qualifying customer already exists.
Referred customers behave differently
Companies consistently find that referred customers stay longer and churn less than customers arriving from paid channels, because a person who vouched for the product pre-selected them.
Longer retention raises the lifetime value the company can expect, which in turn raises what it can justify paying up front to acquire that customer.
The referring customer also tends to stay longer after referring, since having recommended something publicly makes leaving it less likely.
Qualification rules control the risk
Payment almost never triggers at sign-up. Programs require a first purchase above a threshold, a completed billing period, or survival past the return window before the bonus is released.
These conditions exist because sign-ups are cheap to fake. Without a qualifying event, a referral program becomes a mechanism for paying people to create accounts.
The delay also aligns the payout with revenue. The company pays the bonus out of money it has already received rather than in advance of it.
Caps and exclusions keep the program from being farmed
Annual limits on the number of paid referrals stop the program becoming a business for a small number of high-volume participants who are effectively reselling access.
Exclusions for household members, shared addresses and reactivated accounts target the most common ways a single person generates referrals from both sides.
Programs that skip these controls tend to be withdrawn quickly, since abuse costs scale faster than the genuine word-of-mouth the program was built to encourage.
The bonus size signals the growth stage
Unusually large referral payouts generally indicate a company buying growth ahead of profitability, funding acquisition from investment rather than from margin on the product.
As a business matures and its customer base becomes harder to grow cheaply, referral values typically fall or convert into smaller ongoing incentives.
Watching a program shrink is therefore less a sign of stinginess than a sign that the company has shifted from acquiring customers to earning from the ones it has.