Franklin Hugh Money

At its worst, the money was leaving at roughly nineteen hundred dollars an hour. Every hour. Around the clock. Stolen.

This episode is about the year the future of online payments nearly bled to death in Palo Alto, and it teaches three things: what interchange is, which is the toll the card networks collect on every swipe; what a chargeback is, which is how a stolen payment becomes your problem; and unit economics, which is the discipline of asking whether each individual transaction makes money or loses it, before you celebrate how many transactions you have.

Episode three left two companies one block apart on University Avenue. X dot com, Elon Musk's everything bank, with its rented charter and its twenty-dollar signup bonus. And Confinity, run by Peter Thiel and a cryptographer named Max Levchin, whose product, called PayPal, let one person email money to another. Free, instantly, no bank branch involved.

Both companies had discovered the same accelerant: pay people to join. And both had discovered the same customers, because the place person-to-person payments turned out to matter was eBay, the auction site, where millions of strangers needed to pay other strangers and mailing a check took a week.

So through the winter of nineteen ninety-nine into two thousand, two venture-funded companies stood a block apart, handing out cash bonuses, fighting to buy the same users faster than the other one could. Musk has described it since as a race to consume capital, and he was funding a meaningful part of his side with his own fortune. Signups exploded on both sides. So did the burn.

Understand what each free transfer cost, because this is the first teaching object. When a customer funded a payment with a credit card, the card networks and banks took their toll, roughly two to three percent of the transaction, before the money moved an inch. That toll is called interchange. The customer paid nothing. The recipient paid nothing. The company in the middle paid the toll on every single transaction, and charged no one, because growth came first. The product was popular for the same reason it was ruinous. Multiply a small guaranteed loss by a million happy users and you have built a machine that converts venture capital into market share, with no off switch.

In March of two thousand, the two companies did the only sane thing. They stopped burning money at each other and merged, X dot com and Confinity, into one company that kept the X dot com name, with Musk as the largest shareholder. Days later, the merged company closed a financing of roughly one hundred million dollars, and the timing deserves a sentence of respect: the round came together in the exact weeks the Nasdaq peaked and broke. Within a month, the market for funding money-losing internet companies was gone. They got the money over the wall as the wall was collapsing. Musk has called it luck. It was also the difference between everything that followed and nothing.

Because what followed was the fraud.

A payment service that moves money instantly, worldwide, on a credit card number, is the most attractive machine ever built, from the point of view of a thief. Organized fraud rings, prominently including groups operating from Russia, discovered the merged company at scale in two thousand. The mechanics are the second teaching object, and they still govern every dispute on your own card today.

A thief pays with a stolen card number. The money moves, and the goods move. Weeks later, the card's real owner sees the charge and disputes it. The card network reverses the payment, and here is the part that matters: the reversal does not land on the thief, who is gone, and it does not land on the cardholder, who is protected. It lands on the company that accepted the payment, as a document called a chargeback. The chargeback claws back the full amount, adds a penalty fee, and, if a company accumulates too many of them, the card networks can raise its costs or cut it off entirely, which for a payments company is death.

So the merged company was paying interchange on every legitimate transaction, and eating the full loss plus penalties on every fraudulent one. By the middle of two thousand the fraud losses were running at millions of dollars a month, against revenue that barely existed, because most of the service was still free. That is the arithmetic behind the nineteen hundred dollars an hour. The company had ten million users, top-ten-website traffic, and a business that lost money faster the more people loved it.

Growth had been the strategy, and growth was the murder weapon. That sentence is the whole discipline of unit economics: volume multiplies whatever the underlying transaction earns, including a negative number.

Two things saved it, and neither one was a slogan.

The first was engineering. Levchin and his team built software that studied the patterns of fraud, the way accounts linked to each other, the way stolen money hopped between them, and flagged the networks of theft a human investigator could then unravel. They named the system Igor, after a Russian fraudster it helped catch. The same team invented a now-universal test: distorted letters a human can read and a machine cannot, so a program cannot open ten thousand accounts by itself. If you have ever squinted at wavy text to prove you are human, you have used a fraud tool this company built to stop the bleeding. Fraud losses fell from catastrophic toward manageable, and the company's ability to detect theft became, in Thiel's telling and in the record, a genuine competitive weapon.

The second was the end of free. The company began charging merchants, the sellers on the receiving end, a fee per transaction, in the same neighborhood as the interchange it was paying out. The customers stayed. Revenue went from almost nothing toward tens of millions a year with startling speed. The product had been worth paying for all along; the war had just made everyone afraid to ask.

Now the audit, because the legend here is subtle. The legend remembers PayPal as a brilliant idea that conquered the internet. The record shows a business that was dying of its own popularity in the middle of two thousand, rescued by a fraud-detection lab, a fee schedule, and a hundred-million-dollar financing that beat the closing window by days. The idea was the same all year. The unit economics changed, and that is the entire difference between the company that died in the textbooks and the company that lives in them.

Carry the tool with you. When any business, a startup, a delivery app, a bank with a bonus, grows loudly, the question that cuts through every press release is the one this company had to answer at nineteen hundred dollars an hour: what does one transaction earn, after everyone in the middle is paid, and after the thieves take their share? Loss reserves and fraud costs sit in plain sight in the filings of every payments and lending company on earth. Read those before you admire the growth.

One more thing was true by the end of two thousand, and it sets up everything. The company had been rescued, but its chief executive had not been there to see it. He was gone by October, removed while he was on an airplane over the Pacific, by the people he had just merged with.

The largest shareholder of the company. Fired. In absentia. On his honeymoon.

Who has the power to do that, and where exactly that power is written down, is episode five. The Letters.