What is a refer-a-friend program and how does it work?

How does a refer-a-friend program work step by step?

The customer receives a unique link or code, shares it, and the recipient uses it when signing up. The system attributes the new customer to the referrer, waits for the qualifying action, then releases rewards to one or both sides.

What should count as a qualifying referral?

Four models are common: signup, first purchase, a spend threshold, or sustained activity such as a maintained balance or repeat order. The later the qualification, the lower the fraud and the higher the customer quality, at the cost of slower, less motivating payouts.

• Signup: fastest, highest fraud exposure, weakest quality signal
• First purchase: the common default; balances speed against quality
• Spend threshold: filters low-value acquisitions
• Sustained activity: strongest quality signal, slowest to reward

Should both sides be rewarded?

Usually yes. A double-sided reward gives the advocate something to offer, not something to extract, which makes the ask far easier socially. Single-sided programs pay less but ask the advocate to promote a product with no benefit to the person they are recommending.

Where should the refer-a-friend prompt appear?

After a moment of demonstrated value (a delivered order, a completed task, a positive support resolution), not buried in account settings. Placement moves participation more than reward size does, because the ask only works when the product has just proved itself.

How long should a referral link stay valid?

Long enough to cover a realistic decision cycle, short enough to bound your liability. Consumer programs commonly use 30 to 90 days from click; longer windows create attribution disputes, and links left valid indefinitely accumulate obligations you cannot forecast.

Should refer-a-friend rewards be capped?

Cap the referrer, not the referred person. An annual or per-period cap on earnings bounds fraud exposure without ever turning away a new customer. That is the asymmetry Toss uses: the inviter caps out at twenty rewards while every invitee still gets paid.

The common refer-a-friend mistakes

Rewarding signup instead of a meaningful action, hiding the program in settings, setting a reward that exceeds the margin it generates, and launching with no written terms, which leaves no basis to withhold a reward when abuse appears.

• Qualifying on signup, which is cheap to fake and predicts little
• Burying the prompt where no one encounters it
• A reward larger than the gross margin it produces
• No published terms, so abuse cannot be challenged

How do you stop people referring themselves?

Combine controls rather than relying on one: distinct payment instrument, distinct device, distinct delivery address where relevant, and a holding period before the reward is released. ShopBack publishes its device rule openly, which deters the behaviour up front instead of only catching it afterwards.