Viral Loop

by ADAM L PENENBERG

  It was only logical that Confinity drop everything to concentrate on auctions. “We would build PayPal’s payments network on top of eBay’s marketplace,” declared Eric M. Jackson, a Confinity marketing executive and author of The PayPal Wars: Battles with eBay, the Media, the Mafia, and the Rest of Planet Earth. This would be the first time a viral network would be stacked atop another viral network, and it offered a significant advantage to going it alone. Everything hinged on PayPal’s viral adoption, but users were spread all over the world. A clump of customers formed here, another there, a third another place, but they existed as separate satellites lacking synergy. On the other hand eBay offered a single destination, with users clustered in one space and, on average, processing multiple transactions with different buyers each day.

  In the way that there was little sense in selling on eBay if there were no buyers (and, conversely, little incentive to shop if there was nothing to buy), sellers had dual reasons to promote PayPal. It helped streamline business over the Internet and the seller profited from it, multiplying the effect of Paypal’s $10 referral fee. In essence, PayPal became a product to hawk on eBay just like any other. To help market it, engineers created a clickable HTML logo that eBay sellers could insert into their auctions, thus closing the viral loop. Not only were sellers pushing PayPal; buyers would, too. “As sellers used their listings to train buyers to look for PayPal, buyers would in turn learn to ask sellers who weren’t using PayPal if they could pay with our service,” Jackson wrote. “PayPal customers in eBay’s community would recruit other users for us—it was viral growth in its most basic form.” PayPal could then piggyback on the auction portal as it continued on its merry way to hundreds of millions of users, one viral network superimposed over another.

  It didn’t take a PhD in mathematics to see the possibilities. A month after doubling down on eBay, PayPal counted the number of auction logos it had on the site and calculated it was accepted on 6 percent of all listings on eBay, up from 1 percent the previous month. At the same time PayPal’s user growth began to mirror the classic hockey-stick curve indicating exponential growth. On February 3, 2000, it tallied one hundred thousand registered users, and by mid-April, 1 million. But PayPal wasn’t alone in chasing after online gold. Competitors like Dotbank, which launched with venture capital around the same time, touted similar services and a snazzier-looking website, while X.com sought to be a one-stop-shop for financial services.

  A third, PayMe, was led by the CEO of a firm that had funded Confinity; this looked to Thiel like an investor partner had used his due diligence to back a rival. Thiel demanded he either sell back his PayPal stake or pull the plug on PayMe, which he rejected in favor of remaining a much despised shareholder. They all duked it out over who could sign up the most users, with Dotbank matching PayPal’s referral fee while X.com broke the bank by gifting $20 per new user, and kept tabs on one another, often copying services and strategies. Finally, PayPal raced headlong into auctions and only X.com followed.

  As they both grew, the “burn rate”—negative cash flow—reached epic proportions, with every new user forcing them deeper into debt. Thiel believed the one to win would be the first to file for an IPO. He calculated that X.com with its name brand management team had a slightly better chance than Confinity. X.com’s CEO was Elon Musk, a young entrepreneur born and raised in South Africa who was fresh off having sold a company to AltaVista for more than $300 million.

  [ MERGING, BUT NOT OF THE MINDS ]

  The two companies merged in March 2000, with Bill Harris, a former CEO of Intuit, leading the combined company and Musk, the largest single shareholder, becoming chairman. Thiel was the chief financial officer and went off to the capital markets to raise more money. They agreed that Confinity would drop its name and become PayPal. It would follow its present course on eBay while also being used to sell its users on X.com’s premium financial services, which on the surface seemed a nice synergy. To stop the hemorrhaging, X.com ended its $20 referral fee and PayPal sliced its fee in half to $5. The effect on X.com was immediate: its growth stalled, while PayPal continued growing at a frantic pace with barely a ripple.

  But, as often happens in mergers, the two companies clashed, with X.com’s button-down culture running headlong into PayPal’s freewheeling environment. Perhaps telling, it took two full months before email from PayPal could be delivered to X.com users—and they were a technology company. The three-headed hydra of Harris, Musk, and Thiel couldn’t agree on strategy. Harris, who came of age in the old economy, proposed charging users to send money, since PayPal’s original idea of living off the float wasn’t working; users spent money as fast as it came into their accounts. Thiel, however, vehemently opposed this because not only would it stall the company’s viral growth, it could drive them out of business. Their disagreements escalated when Thiel learned that Harris had donated $25,000 of company money to the Democratic Party. After a heated exchange, Harris accused Thiel of insubordination and Thiel, who had just raised $100 million in the financial markets, quit in disgust. The real tension, however, arose between the two engineering staffs. Musk wanted to transition to Microsoft NT servers to relaunch the combined site, while Levchin was adamant about sticking with Unix.

  Despite its brisk pace of user acquisition (half a million new accounts a month and climbing), the combined company continued to bleed red, unable to fix on any one course. Its infrastructure was overtaxed, the website becoming less reliable under the swarm of new users, and customer service was unable to handle the hurricane of calls that came into the company’s switchboard. It got so bad that employees communicated by cell phone, since they couldn’t get an outside line anymore. Six days after Thiel’s resignation, Musk, responding to concerns expressed by the rank and file, called an emergency board meeting to replace Harris as CEO. Musk was named the new CEO and Thiel returned as chairman, with the board commanding the new management team to solve its cash flow problems. Its margins were negative 3.5 percent because it paid $2.50 for every credit card transaction plus another $1.04 to fraud on every $100 in payments. At this $10 million a month burn rate, the $100 million Thiel raised wouldn’t last long.

  Meanwhile Levchin was infuriated by Musk’s order to transition to Microsoft servers and throw all of his engineering resources into a revamped website while fraud threatened to run them out of business. With Thiel out of the picture, he had resigned himself to the inevitability of the platform change away from Unix. Levchin thought about quitting and bided his time by creating a program to compare Unix to Windows that showed that Microsoft had 1 percent of the scalability of Unix. Then he became interested in looking into the fraud problem, and what he discovered frightened him. While the number of transactions was increasing at a fierce rate, fraud was going up even faster. It was only about 1 percent of company revenue, but when Levchin charted out its growth, he could see that, left unchecked, it was on its way to 5 percent, then 10 percent. By midsummer PayPal was losing $10 million a month to fraudsters, many of them backed by the Russian mob. When he emailed Musk with his findings, telling him the migration to Microsoft and site redesign would have to wait until they did something about fraud, the CEO brushed it aside.

  But the final straw was when Musk took steps to phase out the PayPal brand in favor of X.com, a domain name that, despite costing him $1 million, company surveys showed the public had little appetite for, since it was reminiscent of the porn industry. Just before leaving for the Sydney Olympics, Musk ordered the PayPal name expunged from the website. Levchin gathered his former executive team together and led a PayPal revolt. They took up a petition to demand that the board remove Musk or they would quit. The coup was successful, with the board removing Musk as CEO while he was on a flight to Australia—“That’s the problem with vacations,” he would later remark—and appointing Thiel as interim CEO, a title he would end up keeping.

  Musk, who would later go on to found Tesla Motors, a builder of electric cars, and Space X, which seeks t
o colonize space, was understandably bitter, blaming his decision to migrate the company’s infrastructure over from Unix to Microsoft for the shabby treatment, as well as simply having a different point of view. He told Fortune, “Peter, Max, and I are not directly aligned philosophically. Peter’s philosophy is pretty odd. It’s not normal. He’s a contrarian from an investing standpoint and thinks a lot about the singularity. I’m much less excited about that. I’m pro-human.”

  [ FRAUD FIGHTER ]

  Thiel’s first step after wresting control of PayPal was to turn Levchin loose on the fraud problem. The more the company scaled—the more accounts were opened and the more money that coursed through its network—the more bogus charges scaled along with it. In the middle of 2000 one criminal ring stole, over a four-month period, $5.7 million by creating a robotic script to automatically open hundreds of thousands of PayPal accounts at a time. With filched credit cards the ring could send payments through seemingly endless layers of fraudulent accounts until transferring the balance out at the end. Instead of receiving thousands of Beanie Babies or used books and compact disks, the criminals siphoned millions in cold hard cash. Somebody had to pay for it. When a credit card user contests a charge, the merchant is the one left holding the bag, which in this case was PayPal. Around the same time MasterCard added to PayPal’s woes by fining the company $313,600 for excessive charge-backs. At fraud’s peak, PayPal was losing an average of $10 million a month.

  Working with PayPal engineer David Gausebeck, Levchin’s first step was to address fake accounts. Whatever system they created would have to complicate the registration process just enough to hamstring automated methods of creating accounts while not driving away legitimate users and disrupting PayPal’s viral growth. He realized the key lay in the difference between a machine and a human. In 1950, Alan Turing of the Computing Laboratory in Manchester University created a blind test to demonstrate whether a machine could exhibit human-like intelligence. A man would sit at a keyboard and, without being able to see or hear whom he was communicating with, type out a conversation. If after a time he was convinced he was communicating with a human only to learn that it was a computer, the machine would be deemed to exhibit intelligence. Levchin realized he needed to design and implement a “reverse Turing test.”

  The solution was ingeniously simple. Levchin simply added a slight wrinkle to the sign-up process. A would-be registrant was instructed to retype into a separate box a random series of hazy letters that sat atop a yellow background etched with slender black lines in a grid—something a person could easily manage but a machine couldn’t because it wouldn’t be able to decipher the symbols through the slight distortion. Just for fun, Levchin inserted an “Easter egg,” a joke or secret message that programmers sometimes leave behind, which he intended for one of his arch fraud nemeses, a Russian who had set up thousands of accounts under the name Gregory Stivinson. Levchin called his code “GSWeb” and when anyone tried to screw with the system, it would taunt back by spelling out Nice try. Shortly after instituting the Gausebeck-Levchin test the Russian cybercriminal who inspired it sent Levchin an angry email, cursing the Ukranian-born programmer for taking away his livelihood. Levchin replied in kind: “I have a great job for you, which involves wearing an or-ange-colored jumpsuit.” Shortly after, Levchin’s reverse Turing test became a staple in online commerce.

  But Levchin was only half done. Now that he had found a way to prevent crime rings from faking accounts on a large scale he needed to stop criminals who were already inside or able to break into PayPal’s system. This he accomplished by crunching an algorithm that would drive a software program he named Igor, after another Russian nemesis, to automatically flag suspicious activity and freeze the accounts. Levchin accomplished this by visually mapping out legitimate and illegitimate financial transfers and looking for disparate properties. A human investigator would then review the transactions, contact the registrant, and decide whether to unfreeze the funds. “If you reduce [fraud] down to essentially a picture, you can combine the aggregate suspiciousness into a single image,” Levchin says. “Symmetry turns out to be a fantastic predictor of whether something is fraud or not, and you can basically reduce investigation time from several days per scam to several seconds.”

  Of all the illicit transactions PayPal registered, 75 percent arose from bogus credit cards while the remainder was merchant fraud. After Levchin unleashed Igor, credit card fraud dropped to less than 1 percent.

  [ EBAY STRIKES BACK ]

  Fraud was but one problem plaguing PayPal. It also had to deal with a slew of new competitors, with names like PayPlace and gMoney, eMoneyMail and Yahoo PayDirect, all vying for a slice of the online transactions market. PayPal’s most direct threat, however, came from eBay, which viewed it as a parasite sucking money out of its network. Meg Whitman had long wanted to close the loop on transactions and profit from each one. In 1998 eBay purchased Billpoint, a credit-card processing business, which it promptly shut down so it could be rebuilt to handle the kind of volume millions of auctions would bring. As a public company, eBay had to find adequate strategies to manage the kind of fraud that would likely ensue before it dared to release a product. Then came the severe outages that forced the company to focus exclusively on its infrastructure until it became robust enough to sustain its stratospheric growth. Meanwhile PayPal continued to grow. By April 2000 approximately 25 percent of all eBay listings involved PayPal, double the level in March.

  Since going public, eBay had moved far from the people-are-basically-good, hippy-dippy ethos with which its founder, Pierre Omidyar, had started. Now it was a cold and calculating corporation with merely the patina of down-home folksiness. EBay’s first shot over the bow involved the auction logos promoting PayPal. The new regulation, which eBay claimed was intended to limit distracting third-party ads, mandated logos a quarter the size of PayPal’s. Every auctioneer with a PayPal logo had to resize it to meet the new regulation.

  EBay was laying the foundation for Billpoint, whose partners included Wells Fargo (handling the back end) and Visa. Billpoint was rolled out in April 2000, with eBay announcing it with large clickable banners that would take users through the registration process. EBay also ensured that its Billpoint logos would have far greater visibility by placing them at the top of auctions, while PayPal’s would be relegated to the bottom. Then to further encourage sign-ups, eBay offered a free-listing day so any item a seller put up for auction would not cost the usual 25 to 50 cents. In one day Billpoint jumped from a 1 percent to a 10 percent share of auctions. “To put this gain into perspective, this free-listing day propelled Billpoint to a listing share level that it took PayPal a month to reach after we began focusing our efforts on eBay,” PayPal Wars author Jackson wrote. “It was utterly demoralizing.”

  Then a strange thing happened. Billpoint’s share slipped to 6 percent over the course of several days. Sellers, taking advantage of the free offer, had flooded their auctions with new items, which sat around longer than usual before being sold. In the interim PayPal continued its surge, gaining a 5 percent share of listings, from 25 to 30 percent. EBay tried again with another free-listing day, with similar results. Billpoint topped out at 14 percent before free-falling to 7 percent. A third free-listing day did little to improve Billpoint’s share. Because PayPal continued to grow virally during this spate of free-listing days, which cost eBay millions in lost revenue, it actually increased its lead over Billpoint—from a spread of 20 percent in April to 26 percent in May.

  EBay tried another tactic. It introduced Buy It Now, a feature that would enable a seller to fix a price rather than take bids and made Billpoint the default payment option. To use PayPal a seller would have to change the settings, an additional step that could drastically slow PayPal’s adoption. This time it was eBay’s own community that rose up against the new policy, generating a fair amount of negative press coverage. Soon PayPal was grabbing 70 percent of all transactions on eBay. Clearly it had achieved a poin
t of nondisplacement, and nothing eBay did could knock it off its payment pedestal. Being viral, it grew simply because it had already grown. Simultaneously PayPal transitioned users away from credit cards, which cost PayPal a 2+ percent fee on every transaction, and to interest-bearing bank accounts, and began levying a 1.2 percent fee on sales, which was still far less than Visa charged. Along with a vast reduction in fraud, thanks to Levchin, PayPal’s burn rate was slashed. Now, as the company grew, its financial situation improved along with it. While still deeply in debt, PayPal was on the road to profitability, while its competitors crashed and burned. The construction of a call center in Omaha, Nebraska, to handle customer service also alleviated much of the bad press the company had received.

  [“IF IT’S WAR YOU WANT, IT’S WAR YOU’LL GET” ]

  With the company on its way to 200 million PayPal accounts worldwide, Thiel and Hoffman agreed it was time to seek a buyer, set their minimum price at $600 million, and approached Google and Yahoo. Both claimed PayPal’s reliance on eBay (70 percent of its business took place there) made it too big a risk. What if eBay blocked PayPal? Then a $600 million investment would be worthless. Of course, if eBay did that it would risk running afoul of antitrust statutes, but for Thiel and Hoffman it would be a bullet from the grave; by the time it could be litigated PayPal would be dead. Thiel and Hoffman agreed it was time to prepare for an IPO, even though the economy was mired in recession.