Double-Sided Referral Program Statistics: 89% Pay Both

Double-sided referral program statistics are hard to find. Most platforms won't publish reward structures across live programs, so growth teams end up copying whatever a competitor does and hoping it's representative. ReferralFlo's State of Referral Marketing benchmark reviewed 35 currently running, publicly verifiable referral programs and found that 31 of them, 89%, reward both the referrer and the friend they invite. The other four are deliberate exceptions, not oversights. That distinction matters more than the headline number.

How we verified these double-sided referral program statistics

A program counted as verifiable only if its referral terms were published on a live, dated page: reward amount, currency or credit type, and any qualifying condition such as a closed deal or first order. Programs hidden behind a login wall, or mentioned only in a press release, were excluded from the count. No self-reported survey data made the cut.

That screen is strict on purpose. A press release can describe a program that never shipped, or describe terms that changed the following quarter. A published, dated terms page is the closest thing to a verifiable snapshot. The 35-program set spans SaaS, fintech, e-commerce, and consumer banking, each with terms current as of the review date.

This is a snapshot, not a permanent ranking. Terms change, and a benchmark taken today can read differently in six months; for more on why that happens, see why referral program benchmarks go stale fast.

The 89% figure and its four exceptions

Four of the 35 programs pay only one side, and each does it deliberately rather than by accident. Zerodha, the Indian discount broker, publishes a referrer-only reward tied to brokerage the referred account generates. GXS, the Singapore digital bank, runs a one-sided structure too. Neither omission looks like an oversight; both sit inside regulated, transaction-linked financial products where a cash reward for the new customer changes the compliance picture.

Under the FTC's endorsement guidelines, a material connection between a recommender and a reward must be disclosed to the person receiving the recommendation. That disclosure is straightforward for a SaaS trial or a discount code. It gets harder for a brokerage account, where a cash incentive to the new customer can start to look like an inducement rather than a referral courtesy. Zerodha publishes its referrer-only terms directly, which is exactly the kind of transparency this review required to count a program as verifiable.

The pattern holds across the exceptions: regulated category, referrer-only reward, published terms. For the compliance mechanics behind that choice, see how to launch a fintech referral program that clears compliance review. Outside regulated finance, none of the 35 programs reviewed skipped the friend's reward.

The symmetry benchmark: equal rewards beat overweighting the friend

Among the 31 double-sided programs with comparable, like-for-like reward pairs, 10 of 16 pay the referrer and the friend an identical amount or percentage. That runs against the common advice to weight the reward toward the new customer to offset their higher perceived risk. In this two-sided referral rewards data, symmetry was the more frequent design choice, not the exception.

That's a real finding, but it isn't a mandate. A Series-A B2B SaaS company giving both sides a $50 credit and a DTC beauty brand giving both sides 20% off are typical of the equal-split pattern seen in the sample. Six of the 16 comparable pairs still skew toward the friend, usually where the referrer's existing loyalty tier already carries its own perks.

These referral reward benchmarks describe what programs actually publish, not what converts best for any single business. Before copying either pattern, run your own numbers. On a $60 average order, a matched $12-and-$12 reward is a very different cost than a $12-and-$25 split, even though both are "double-sided." The reward calculator works out the maximum sustainable double-sided reward from your margin, so the split becomes a decision instead of a guess.

What this means for your reward design

The prevalence data supports a default, not a mandate: start with an equal, double-sided reward unless a regulatory constraint or a margin ceiling says otherwise. Most of the market does the same, and the four exceptions in this sample all had a specific, disclosed reason for departing from it.

Model the payout before launch. Run it through ReferralFlo's referral management software to set escrow conditions, so a reward only releases once a deal closes or a first order ships. Then revisit the split once you have real redemption data instead of launch-day assumptions. Reward economics shift as your average order value or contract size shifts, so a split that made sense at seed stage can be wrong two years later.

For the mechanics of building a reward that clears both a finance review and a customer's expectations, our guide to double-sided reward design covers the structure this post only measures.