A growth loop, by definition, is something where using a product generates more users of that product by itself, without a marketing team paying to send more users every time it occurs. That’s it. That’s the whole definition. Everything else is elaboration.
Normal marketing is linear. You pay for a customer via an advertisement, the advertisement works, and you get a customer. When you stop paying for customers, they stop coming. Loops don’t work like that. Once they’re spinning up, they require no ongoing investment because paying/rewarding the user for performing an action is what brings the next user in line, rather than spending from an external marketing budget.
PayPal is the classic example here mostly because its implementation of the concept was so primitive that there’s literally no way to misunderstand it. In 1900, PayPal paid new users $20 when they signed up, and another $20 when anyone they referred signed up. That new user, now with $20 in PayPal, had every incentive to refer a friend of their own. In stretches, that compounded to growth rates of 7% to 10% per day in the company’s early life. (These numbers are from Peter Thiel’s book but are echoed elsewhere) Eventually reaching over 100 million users, according to most accounts of the company’s founding, despite the payout being slowly reduced from $20 down to $10, down to $5 to prevent the whole thing’s economics from exploding out of control. Years later, Peter Thiel estimated PayPal probably spent $60 million to $70 million on the program during its first couple of years of existence.
In order for a growth loop to occur, three specific things have to happen, in order, over and over: A user does something, that action results in something occurring that is visible or valuable to another potential user, who then becomes a new user who does that same action. Strip out any one part, and it’s not a growth loop; it’s a coupon.
That’s what makes growth loops different from standard referral programs. Any referral program that quietly evaporates when you stop encouraging people to use it was never actually a loop. It was just a marketing campaign with a loop’s bright, colorful jersey pulled over it. PayPal didn’t work because you were given an “invite a friend” option that was tacked onto the existing product. You got paid the fastest by inviting a friend. Sending invitations was the primary way you used PayPal. The mechanic and the incentive were one and the same action.
Lesson two: Somewhere inside those PayPal numbers is where everyone screws up. A growth loop doesn’t cost less money; it just shifts where the money goes. Instead of paying for Facebook or Google to send customers your way, PayPal was paying its users directly. With literal cash. And as mentioned above, it had to steadily reduce that payout as the company grew, because expanding at a rate of 7% to 10% per day is fantastic until you send your CFO to torture prison after they crunch the numbers.
Bonus tips, if you’re actually trying to build one of these.
-The reward can’t be just anything. It must be something your product can uniquely provide. PayPal could give people money because transferring money was PayPal’s business. Arbitrary amounts of money coming from nowhere from a company that doesn’t move money would be insane. Some other cost problem entirely.
-The action you perform to send that reward to another user can’t feel like an interruption to the product itself. No one at PayPal needed convincing to refer a friend. Friend referrals were literally how you got money out the fastest.
A growth loop doesn’t create value; it multiplies it. PayPal worked because email-to-email money transfers were absolutely worth $20 at the time; frankly, they still are. If people love your product, they will tell other people because they love your product. If they don’t, a kickback campaign will assemble users who love coupons just as eagerly.
Ordinary marketing is reaching outside the user base and pushing people toward a product. A growth loop is giving users of your product a tiny, automatic reason to turn that initial user into two.
