Dropbox grew from 100,000 users to 4 million in about 15 months by giving away free storage space to anyone who referred a friend, a roughly 3,900% increase with no traditional advertising spend, according to Viral Loops. The program hit a viral coefficient of 0.35, meaning every 10 users brought in 3.5 more, and that compounding is what a paid acquisition budget cannot buy: growth that gets cheaper as it scales instead of more expensive.
Most founders hear that number and build a "refer a friend" button. That is not what Dropbox built. It built a loop with three specific properties, and the companies that replicated the result copied those properties, not the button.
The incentive has to run both ways
Dropbox rewarded the referrer and the new user with free storage. PayPal, earlier, rewarded both sides in cash: $20 to the new signup and $20 to the referrer, a program that cost an estimated $60 million and took the company from 1 million to 5 million users between March and September 2000, according to Viral Loops. Airbnb's rebuilt referral program, relaunched in January 2014, gave both the referrer and the invited friend $25 in travel credit and produced a 300% increase in daily bookings and signups, per Airbnb's own engineering account. In every case, only rewarding the referrer was rejected. A one-sided incentive asks someone to vouch for you in exchange for a kickback, which reads as a sales pitch. A two-sided incentive gives the new user a reason to say yes on the spot, which is what actually converts a share into a signup.
The Airbnb detail worth sitting with: that 300% lift came from a five-person team working for three months, not a redesign of the incentive amount. The team's real work was rebuilding when and how the ask appeared in the product, moving it to the moments when a guest had just had a good trip, not burying it in a settings menu. The incentive was the same shape PayPal and Dropbox used. The placement was the leverage.
Timing beats size
Harry's proved this before the company had a product to ship. Its pre-launch "Milestone Referral" campaign used tiered rewards, better prizes at higher referral counts, and collected over 100,000 email signups in one week, with 77% of those signups coming from referrals, according to the Tim Ferriss Blog account of the campaign. There was no product yet, no storage to give away, no cash reward. What Harry's had was urgency: a fixed pre-launch window and a visible leaderboard of who had referred the most people. The lesson generalizes past pre-launch: a referral ask with a deadline or a milestone outperforms an ask that sits in an account settings page indefinitely.
The underlying reason any of this works is not mysterious. Nielsen's Global Trust in Advertising report, based on a survey of more than 28,000 internet respondents across 56 countries, found that 92% of consumers trust recommendations from friends and family above every other form of advertising, up 18 percentage points since 2007, according to Nielsen. A referral loop is not a growth hack bolted onto a product. It is the only acquisition channel that starts with trust already established, which is why it converts at a rate paid channels cannot match.
None of this makes a referral program permanent. Tesla ended its long-running customer referral program worldwide on April 30, 2024, discontinuing perks like free Supercharging miles that owners had earned for years by referring buyers, according to Electrek. Tesla had already narrowed the program once before, cutting Supercharging-mile rewards on vehicle and solar purchases as of September 18, 2021 and limiting the remaining reward to a $500 credit on Solar Roof referrals only, per Tesla's own program terms. A referral program is a marketing line item with unit economics, and those economics change as a company's growth needs and margins change.
The action for this week: if you are building a referral loop, write down the two-sided incentive, the exact moment in the product where the ask will appear, and a number, cost per referred customer versus cost per paid customer, that you will check monthly. The day that number flips, you will know to redesign the program or sunset it, instead of finding out the way most companies do: after the budget owner cuts it with no data to defend keeping it.