Referral Program Design for Apps: Rewards That Stay Fair

What makes a good referral program design for an app?

A good app referral program rewards the inviter and usually the friend, pays only after the friend completes a meaningful action in your app, limits how much one account can earn, and tracks every invite with a link plus a code that still works when the link breaks. The reward should be something your product already gives out, like in-app currency, storage or a free month, so you control its cost. Then you judge it by what a retained referred user costs you, not by how many people signed up.

This guide covers the design decisions for apps, games and rewards sites. The fraud side is covered in How to prevent referral fraud.

One-sided, double-sided and other reward structures

The first choice is who gets rewarded. A one-sided program rewards only the inviter. A double-sided program rewards both people, which gives the friend a reason to accept and gives the inviter a better pitch than “help me earn something”.

A well documented double-sided example is Dropbox. Its help center says Basic accounts get 500 MB per referral, up to 16 GB, and Plus, Family and Professional accounts get 1 GB per referral, up to 32 GB, while the friend gets 500 MB. The reward is the product itself, it is capped, and it completes only after the friend does something real (Dropbox: earn space by referring friends, checked on October 22, 2026).

Fixed bonus or a percentage of the friend’s earnings

For a typical app or game, a fixed bonus is the simpler choice. You know what each qualified friend costs, and users understand it instantly. Price it in your own currency and check it against what a new user is worth to you; CAC vs CPA vs LTV shows how to set that ceiling, and the LTV calculator helps with the math.

Rewards sites and apps where users earn currency often prefer a percentage: the inviter receives, say, 5% to 10% of the coins their friend earns. Done right, it funds itself, because the base is earnings that already brought you revenue. Done wrong, it becomes a fraud magnet. Four rules keep it safe:

• Pay on top, never out of the friend’s balance. The friend should earn exactly what they would without a referrer, or the program punishes the people you want to attract.
• Use confirmed earnings as the base. Count only credits that passed your hold period, and reverse the referral bonus when the friend’s credit is reversed. Reward hold periods explains how to run pending and available balances.
• Pay one level only. Exclude referral commission from the base, so nobody earns on the friends of their friends. Multi-level schemes reward account rings more than real advocates.
• Choose lifetime or time-limited on purpose. Lifetime is generous and a strong pitch; 6 or 12 months limits your long-term cost. Write the choice into your terms either way.

Many programs combine both: a small fixed welcome bonus for the friend after a first qualifying action, plus a percentage for the inviter.

When to pay: after a meaningful action, not on sign-up

Paying on sign-up is the most expensive mistake: a fake sign-up takes seconds, so the reward invites people to refer themselves. Pick a qualifying action that a real user reaches naturally and a fake account finds costly:

• Utility and productivity apps: finish onboarding and use the core feature, for example create a first project or sync a device. Dropbox requires the friend to install and sign in to the desktop app and verify their email before the referral completes.
• Games: reach a level that takes real play, such as the end of the tutorial plus a few sessions on different days.
• Subscription apps: the first paid period after a trial, not the start of the trial.
• Rewards sites: a first confirmed reward from offers or surveys, past the hold period, plus a verified email or phone.

Show both users what still has to happen, such as “Joined, waiting for first reward”, so the inviter nudges the friend toward the action you care about. Set a deadline too, such as 30 days to qualify.

Caps, tiers and limits that keep it affordable

Every reward needs a ceiling, because the accounts that refer the most are disproportionately likely to be farms. Useful limits:

• A lifetime cap per inviter on fixed bonuses, as Dropbox does with 16 GB and 32 GB.
• A rate limit, such as a maximum number of qualified referrals per day or week, with anything above it held for review instead of rejected.
• A cap per friend on percentage programs, for example a maximum commission per referred user per month.
• Modest tiers. Milestones such as a badge at 3 friends and a bonus at 10 reward real advocates. A huge prize at 100 friends mostly attracts people who can create 100 accounts.

Keep referral rewards inside your virtual economy budget. A referral bonus is a currency source like any daily bonus, and it can inflate prices if it is too generous.

Attribution: links, codes and deferred deep links

Attribution decides which inviter gets credit, and it is where programs silently lose referrals. Use more than one method:

On Android, Google’s documentation says the Play Install Referrer API lets your app retrieve the referrer URL of the installed package, plus timestamps for the referrer click and the install start. Put the inviter’s code in the referrer parameter of your Play Store link, read it on first launch, and send it to your server with the new account.

If your app used Firebase Dynamic Links for referral invites, those links no longer work. Firebase’s FAQ says the service shut down on August 25, 2025, and that clicked links return an HTTP 404 response after the shutdown. It points to App Links (Android) and Universal Links (iOS) for apps that only need to open when already installed, and to third-party deep linking providers for full deferred deep linking (Firebase Dynamic Links deprecation FAQ, checked on October 22, 2026). What is a deep link? explains the difference between standard and deferred deep links.

Store rules and disclosure

The store policies below do not ban referral rewards, but they shape them. Read the current text before you launch; these points were checked on October 22, 2026.

• Never reward ratings or reviews. Google Play’s User Ratings, Reviews, and Installs policy says “Don’t offer rewards or incentives for ratings or reviews” and lists asking users to rate your app while offering an incentive as a common violation. Keep any review prompt completely separate from the referral flow. Can you pay for app reviews? covers both stores.
• Do not force store actions. Apple’s App Review Guidelines 3.2.2(x) say apps must not force users to rate, review or download other apps to access functionality, while allowing incentives for specific actions inside the app. A referral reward is optional; never lock features behind inviting friends.
• Contact invites must be individual. Apple guideline 5.1.2(v) says you may not contact people from a user’s Contacts except at the user’s explicit initiative on an individual basis, with no Select All option and no default selection of all contacts, and you must show the user what the message will say and who it appears to come from. The share sheet is the simplest compliant option on both platforms.
• The rewards app itself. If your app pays users to install other apps through an offerwall, Google Play’s rules on incentivized installs also apply. Google Play’s policy on incentivized installs walks through them.

Disclosure matters when users post their links publicly. The FTC’s Endorsement Guides FAQ advises disclosing an incentive when knowing about it would affect the weight or credibility readers give a recommendation, and says advertisers need reasonable programs to train and monitor the people they reward. A referral reward is that kind of incentive. Put a line in your referral terms asking users to say it is a referral link, give them share text that already says so, and never write earning promises into it.

How to measure a referral program

Sign-ups show how attractive the reward is, not whether the program works. Track these, with referred users as their own segment:

• Invite rate: the share of active users who send at least one invite.
• Qualification rate: the share of referred sign-ups that reach the qualifying action.
• K-factor (viral coefficient): invites sent per user multiplied by the share of invites that become qualified users. Below 1, referrals add to your other channels; above 1, each group of users brings a larger one, which is rare and usually short-lived.
• Cost per qualified referred user: all referral rewards paid, divided by qualified friends.
• Retention and revenue of referred users compared with organic users at day 7 and day 30. Rewards app retention covers the loops that keep them.
• Reversal and ban rate of referred users, which is your early warning for abuse.

To see what referrals really add, hide the referral prompt from a small random group for a few weeks; incrementality testing explains how.

Designing against abuse from the start

Most referral abuse is one person playing both roles. The design choices above (a qualifying action, a hold, caps, one level, rewards in your own currency, reversals that follow the friend’s credits) stop much of it before any detection runs. Two more: do not count a friend who signs up from the inviter’s device, and show inviters only a username, a status and their own reward, never the friend’s email or activity.

For device checks, account rings, and what to do when you find a farm, read How to prevent referral fraud and Stop multi-accounting and VPN abuse.

Frequently asked questions