Why Referral Bonuses Exist: The Unit Economics Behind Free Money

Companies do not hand out sign-up bonuses because they like you. They hand them out because you are, from their accounting department’s point of view, an asset they would otherwise have to buy from Google or Meta at a much worse price. Once you see referral programs as a line item on a marketing budget — not a gift — almost everything confusing about them starts making sense: why the rewards exist, why both sides get paid, why the fine print is so fussy, and why every good program eventually gets worse.

This article walks through the actual math. No spreadsheet required, but there will be arithmetic.

Customer acquisition cost: the number that explains everything

Customer acquisition cost, or CAC, is what a company spends, on average, to turn a stranger into a customer. Ad spend, agency fees, landing pages, the salaries of the people running the campaigns — total it all up and divide by new customers acquired. That’s CAC.

For businesses like brokerages, crypto exchanges, and food delivery apps, CAC through paid advertising is brutal. Finance-related keywords have historically been among the most expensive clicks on the internet — search ads in categories like trading and insurance have commonly run tens of dollars per click, and most clicks don’t convert. By the time a paid-ads funnel turns a click into a funded, verified account, industry estimates for all-in acquisition cost in fintech have often landed in the hundreds of dollars per customer.

Now compare the referral channel. A referral bonus is only paid when someone actually signs up and completes a qualifying action. There is no spend on people who scroll past, click and bounce, or sign up and never verify. The company pays for exactly one thing: a completed conversion. If the bonus plus the referrer’s reward costs $30 combined, and the paid-ads alternative costs $300, the referral program isn’t generous — it’s a 90% discount on the same purchase.

That is the whole reason referral programs exist. Everything else is implementation detail.

The lifetime-value math, with a worked example

CAC only matters relative to lifetime value (LTV): the total profit a customer generates over their time with the company. The rule of thumb in growth circles is that LTV should comfortably exceed CAC — a 3:1 ratio is the number people usually recite.

Here’s an illustrative example with round, made-up numbers. Suppose a commission-free brokerage earns revenue from a typical funded account through order-flow arrangements, interest on uninvested cash, margin lending, and premium subscriptions — call it $100 per year per active customer. Suppose the average customer sticks around five years. That’s $500 in lifetime revenue.

Against that $500, paying out a referral reward that usually lands in the single digits is trivially cheap. Robinhood’s gift-stock program is a real-world version of this: the advertised range runs from $5 up to $200, but the odds are heavily weighted so that roughly 98% of participants receive stock at the low end of the range. The headline says $200; the budget assumes $5–$10. That asymmetry isn’t deception exactly — the terms disclose the odds — but it’s a reminder that the “up to” number is a marketing expense ceiling, not an expectation. (Details, and the current terms, live on our Robinhood referral hub.)

Run the numbers from the company’s side: pay ~$10–$20 total per referred account (both sides’ rewards combined), acquire a customer worth ~$500 over their lifetime, and skip the $300 ad-channel alternative. A finance team will approve that program every single quarter — right up until the math changes, which we’ll get to.

Why both sides get paid

Early referral programs often rewarded only the referrer. Modern programs almost always reward both people, and the reason is conversion psychology, not fairness.

A one-sided program turns your friend into a salesperson. “Sign up with my link” reads as “I get paid if you do this,” which is exactly the pitch people have learned to ignore. A two-sided program changes the sentence to “we both get something,” which converts meaningfully better for two reasons:

1. The invitee has a concrete reason to act now. A bonus tied to sign-up creates a reason to stop procrastinating that “this app is pretty good” never will.
2. The referrer feels less like a shill. People share codes far more willingly when they’re handing a friend a discount rather than extracting a commission from them. The social cost of sharing drops, so sharing volume rises.

There’s also a trust-transfer effect: a recommendation from someone you know carries more weight than an ad. Referred customers have generally shown better retention and higher lifetime value than paid-ad customers in study after study — they arrive pre-sold by someone whose judgment they trust. Companies aren’t just buying a sign-up; they’re buying a better class of sign-up.

One side note for referrers: in the US, the FTC’s endorsement rules apply to referral rewards too. If you’re recommending a product and you get compensated when people sign up, that relationship needs to be disclosed — the FTC’s Endorsement Guides spell this out. It’s the same rule this site follows on its disclosure page.

Why programs cap, rotate, and eventually nerf rewards

If referral programs are such a good deal for companies, why do they keep getting worse? Several forces push in the same direction:

Fraud eats the budget. Every referral program attracts self-referral farms: people creating fake accounts to refer themselves, sometimes at industrial scale. Companies respond with qualifying actions (fund the account, complete a trade, verify identity), payout delays, and caps. Binance’s program is a clean example of a structural cap — the fee kickback a new user can receive through a referral tops out at 20%, no matter what a code promises. Anyone advertising more than that is describing a number that cannot exist. Our Binance referral hub covers how that split actually works.

Early cohorts are the best cohorts. When a program launches, the people using it are genuine enthusiasts referring genuine friends. As a program ages and gets listed on every coupon site on the internet (hello), the mix shifts toward bounty hunters who sign up, collect, and leave. The measured LTV of referred customers drops, the LTV-to-CAC ratio compresses, and the finance team starts asking questions.

Growth stage changes the math. A company burning venture capital to grab market share will happily pay top dollar per user. The same company, later, optimizing for profitability, will not. Reward cuts often track a company’s lifecycle more than anything users did.

Campaigns are deliberately temporary. Many programs rotate offers by design — a rich bonus for a launch quarter, a leaner standard rate afterward, different amounts by country. Uber Eats invite discounts, for instance, vary substantially by country and time period; there is no single global “the Uber Eats referral is worth X” number, and anyone quoting one is guessing.

What this means for you, practically

A few durable takeaways fall out of the economics:

• Take bonuses early in a program’s life. Rewards are richest when a company is growth-hungry and the fraud models are naive. Programs get nerfed far more often than they get sweetened. If an offer looks unusually generous, that’s usually a sign it won’t survive the next budget review — not a sign you should wait.
• Assume the realistic case, not the ceiling. “Up to” numbers are budget caps. When a program discloses odds or tiers, the low end is your planning number.
• Enter the code at sign-up, or not at all. On most platforms — Binance is explicit about this — a referral ID cannot be attached to an account after registration. There is no support ticket that fixes it.
• Read the qualifying action. Bonuses trigger on specific events: sometimes account approval, sometimes a deposit, sometimes a trade. Robinhood’s gift stock, for example, triggers on application approval rather than on funding — a friendlier trigger than most, but US-only and SSN-required, because it’s a real brokerage account.
• A bonus never fixes a bad product. If the fees, spreads, or terms are wrong for you, a one-time $10 is a rounding error against years of paying them. Sign up for services you’d want at zero bonus; treat the bonus as a tip.

Referral bonuses are the rare marketing channel where the marketing budget is handed directly to you. Take it with open eyes: know what the company is buying, know what the realistic payout is, and get in before the spreadsheet catches up. For how we check the codes and claims on this site, see how we verify.