میلیراوی

CHAPTER 4

Milliravi

The turmoil in Amazon’s management during the company’s frenzied years of expansion was only the start of a much longer test of faith. In 2000 and 2001, the years commonly thought of as the dot-com bust, investors, the general public, and many of his employees fell out of love with Bezos. Most observers not only dismissed the company’s prospects but also began to doubt its chances of survival. Amazon stock, which since its IPO had moved primarily in one direction—up—topped out at $107 and would head steadily down over the next twenty-one months. It was a stunning fall from grace.

There were several immediate reasons for the stock market’s reversal. The excesses of the dot-com boom had begun to wear on investors. Companies without actual business models were raising hundreds of millions of dollars, rushing to go public, and seeing their stock prices roar into the stratosphere despite unsound financial footing. In March of 2000, a critical cover story in Barron’s pointed out the self-destructive rate at which Web companies like Amazon were burning through their venture capital. The dot-com boom had been built largely on faith that the market would give these young, unprofitable companies plenty of room to mature; the Barron’s story reinforced fears that a day of reckoning was coming. The NASDAQ peaked on March 10, then wobbled and began to spiral downward.

The outbreak of negative sentiment toward Internet companies in general would be nudged along by other events over the course of the next two years, like the collapse of Enron and the 9/11 terrorist attacks. But the underlying reality was that many investors decided to doff their rose-colored glasses and look at Internet companies more pragmatically. And those companies included Amazon.

While other dot-coms merged or perished, Amazon survived through a combination of conviction, improvisation, and luck. Early in 2000, Warren Jenson, the fiscally conservative new chief financial officer from Delta and, before that, the NBC division of General Electric, decided that the company needed a stronger cash position as a hedge against the possibility that nervous suppliers might ask to be paid more quickly for the products Amazon sold. Ruth Porat, co-head of Morgan Stanley’s global-technology group, advised him to tap into the European market, and so in February, Amazon sold $672 million in convertible bonds to overseas investors. This time, with the stock market fluctuating and the global economy tipping into recession, the process wasn’t as easy as the previous fund-raising had been. Amazon was forced to offer a far more generous 6.9 percent interest rate and flexible conversion terms—another sign that times were changing. The deal was completed just a month before the crash of the stock market, after which it became exceedingly difficult for any company to raise money. Without that cushion, Amazon would almost certainly have faced the prospect of insolvency over the next year.

At the same time, rising investor skepticism and the pleadings of nervous senior executives finally convinced Bezos to shift gears. Instead of Get Big Fast, the company adopted a new operating mantra: Get Our House in Order. The watchwords were discipline, efficiency, and eliminating waste. The company had exploded from 1,500 employees in 1998 to 7,600 at the beginning of 2000, and now, even Bezos agreed, it needed to take a breath. The rollout of new product categories slowed, and Amazon shifted its infrastructure to technology based on the free operating system Linux. It also began a concerted effort to improve efficiency in its far-flung distribution centers. “The company got creative because it had to,” says Warren Jenson.

Yet the dot-com collapse took a heavy toll inside the company. Employees had agreed to work tirelessly and sacrifice holidays with their families in exchange for the possibility of fantastic wealth. The cratering stock price cleaved the company in two. Employees who had joined early were still fabulously rich (though they were also exhausted). Many who had joined more recently held stock options that were now worthless.

Even top managers grew disillusioned. Three senior executives recall meeting privately in a conference room that year to write a list of all of Bezos’s successes and failures on a whiteboard. The latter column included Auctions, zShops, the investments in other dot-coms, and most of Amazon’s acquisitions. It was far longer than the first column, which at that time appeared to be limited to books, music, and DVDs. The future of the new toys, tools, and electronics categories was still in question.

But through it all, Bezos never showed anxiety or appeared to worry about the wild swings in public sentiment. “We were all running around the halls with our hair on fire thinking, What are we going to do?” says Mark Britto, a senior vice president. But not Jeff. “I have never seen anyone so calm in the eye of a storm. Ice water runs through his veins,” Britto says.

In the span of the next two turbulent years, Bezos redefined Amazon for the rapidly changing times. During this period, he met with two retailing legends who would focus his attention on the power of everyday low prices. He would start to think differently about conventional advertising and look for a way to mitigate the costs and inconveniences of shipping products through the mail. He would also show what was becoming a characteristic volatility, lashing out at executives who failed to meet his improbably high standards. The Amazon we know today, with all of its attributes and idiosyncrasies, is in many ways a product of the obstacles Bezos and Amazon navigated during the dot-com crash, a response to the widespread lack of faith in the company and its leadership.

In the midst of all this, Bezos burned out many of his top executives and saw a dramatic exodus from the company. But Amazon escaped the downdraft that sucked hundreds of other similarly overcapitalized dot-coms and telecoms to their deaths. He proved a lot of people wrong.

“Up until that point, I had seen Jeff only at one speed, the go-go speed of grow at all costs. I had not seen him drive toward profitability and efficiency,” says Scott Cook, the Intuit founder and an Amazon board member during that time. “Most execs, particularly first-time CEOs who get good at one thing, can only dance what they know how to dance.

“Frankly, I didn’t think he could do it.”

In June of 2000, with Amazon’s stock price headed downward along with the rest of the NASDAQ, Bezos first heard the name Ravi Suria. A native of Madras, India, and the son of a schoolteacher, Suria came to the United States to attend the University of Toledo and earned an MBA from the school of business at Tulane University. At the start of 2000, he was a new and unknown twenty-eight-year-old convertible-bond analyst at the investment bank Lehman Brothers, working in a small office on the fourteenth floor of the World Financial Center.1 By the end of that year, he was one of the most frequently mentioned analysts on Wall Street and the unlikely nemesis of Jeff Bezos and Amazon.

For the first five years of his career, at Paine Webber and then at Lehman, Suria wrote about esoteric subjects like the overcapitalization of telecommunications companies and biotechnology firms. After raising its third high-profile round of debt and losing Joe Galli, its chief operating officer, Amazon demanded Suria’s attention. Working from Amazon’s latest quarterly earnings release, Suria analyzed the heavy losses of the previous holiday season and concluded that the company was in trouble, and in a widely disseminated research report, he predicted doom.

“From a bond perspective, we find the credit extremely weak and deteriorating,” he wrote in what would be the first of several scathing reports on Amazon over the next eight months. Suria said that investors should avoid Amazon debt at all costs and that the company had shown an “exceedingly high degree of ineptitude” in areas like distribution. The haymaker was this: “We believe that the company will run out of cash within the next four quarters, unless it manages to pull another financing rabbit out of its rather magical hat.”

The prediction generated sensational headlines around the world (New York Post: “Analyst Finally Tells the Truth about Dot-Coms”2). Already freaked by the market’s initial decline, investors dropped Amazon, and its stock fell by another 20 percent.

Inside Amazon, Suria’s report hit a nerve. Bill Curry, Amazon’s chief publicist at the time, called the report “hogwash.” Bezos expanded on that assessment when he spoke to the Washington Post, saying that it was “pure unadulterated hogwash.”3

Suria’s analysis was, in the narrowest sense and with the benefit of hindsight, incorrect. With the additional capital from the bond raise in Europe, Amazon had nearly a billion dollars in cash and securities, enough to cover all of its outstanding accounts with suppliers. Moreover, the company’s negative-working-capital model would continue to generate cash from sales to fund its operations. Amazon was also well along in the process of cutting costs.

The real danger for Amazon was that the Lehman report might turn into a self-fulfilling prophecy. If Suria’s predictions spooked suppliers into going on the equivalent of a bank run and demanding immediate payment from Amazon for their products, Amazon’s expenses might rise. If Suria frightened customers and they turned away from Amazon because they believed, from the ubiquitous news coverage, that the Internet was only a fad, Amazon’s revenue growth could go down. Then it really could be in trouble. In other words, the danger for Amazon was that in their wrongness, Suria and other Wall Street bears might prove themselves right. “The most anxiety-inducing thing about it was that the risk was a function of the perception and not the reality,” says Russ Grandinetti, Amazon’s treasurer at the time.

Which is why Amazon’s damage-control response was unusually emphatic. In early summer, Jenson and Grandinetti crisscrossed the United States and Europe, meeting with big suppliers and giving presentations on the financial health of the company. “Even the facts were guilty until proven innocent for a short period of time,” Grandinetti says.

In one trip, Grandinetti and Jenson flew to Nashville to reassure the board of Ingram that Amazon was on sound financial footing. “Look, we believe in you guys. We like what you’re doing,” John Ingram, its president, told the Amazon executives while his mother, Martha Ingram, the company’s chairman, looked on. “But if you go down, we go down. If we’re wrong about you, it’s not ‘oh, shucks.’ We have such a concentration of our receivables from Amazon that we will be in trouble too.”

With Amazon’s reputation and brand getting battered in the media, Bezos began a charm offensive. Suddenly, he was everywhere—on CNBC, in interviews with print journalists, talking to investors—asserting that Suria was incorrect and that Amazon’s fundamentals were fine. At the time, I was the Silicon Valley reporter for Newsweek magazine, and I spoke to both Bezos and Jenson that summer. “The biggest message here is, his cash flow prediction is wrong. It’s just completely wrong,” Bezos told me in the first of our dozen or so conversations over the next decade.

In the transcript of that interview, Bezos seemed, even a decade later, to be full of confidence and conviction, and he was already a steady recycler of tidy Jeffisms. He reaffirmed his commitment to building a lasting company, learning from his mistakes, and developing a brand associated not with books or media but with the “abstract concept of starting with the customer and working backward.”

But when Bezos addressed Suria’s predictions, his comments seemed defensive. “First of all, for anybody who has followed Amazon.com for any length of time, we’ve all seen this movie before,” he said, interjecting cavalcades of laughter between his answers. “I know we live in a period where long term is ten minutes [laugh] but if you take any historical perspective whatsoever… I mean, let me ask you this question. How much do you think our stock is up over the last three years? The stock is up by a factor of twenty! So this is normal. I always say about Amazon.com we don’t seek controversy, but we certainly find it [laugh].”

In fact, times were not normal. The challenge from Suria and the dot-com collapse had changed the financial climate, and Bezos knew it. A few weeks later, Jenson and Bezos sat down to scrutinize Amazon’s balance sheet. They came to the conclusion that even if the company showed reasonable growth, its fixed costs—the distribution centers and salary rolls—were simply too big. They would have to cut even more. Bezos announced in an internal memo that Amazon was “putting a stake in the ground” and would be profitable by the fourth quarter of 2001.4 Jenson said that the company “tried to be realistic about what revenues were going to be and everyone was given a target on expenses.”

But the company couldn’t catch a break in the press. When Amazon announced this goal publicly later in the year, it was subject to a new round of criticism for specifying that it would measure profitability using the pro forma accounting standard—which ignored certain expenses, like the costs of issuing stock options—instead of more conventional accounting methods.

For the next eight months, Ravi Suria continued to pummel Amazon with negative reports. His research became a litmus test for people’s view of the dawning new Internet age. Those who believed in the promise of the Web and who had bet their livelihoods on it were likely to be skeptical of Suria’s negative perspective. But those who felt that the coming wave of changes threatened their businesses, their sense of the natural order, even their identities, were likely to embrace the sentiments of Suria and like-minded analysts and believe that Amazon.com was nothing more than a crazy dream precariously built on an irrationally exuberant stock market.

Perhaps that is why hyperrational Bezos grew so obsessed with the mild-mannered, bespectacled New York analyst. To Bezos, Suria represented a strain of illogical thinking that had infected the broader market: the notion that the Internet revolution and all of the brash reinvention that accompanied it would just go away. According to colleagues from the time, Bezos frequently invoked Suria’s analyses in meetings. An executive in the finance group used Suria’s name to coin a term for a significant mathematical error of a million dollars or more; Bezos loved it and started using it himself.

The word was milliravi.

It is the ambition of every technology company to be worth more than the sum of its parts. It inevitably seeks to offer a set of tools that other companies can use to reach their customers. It wants to become, in the parlance of the industry, a platform.

At the time, Microsoft was the archetype for such a strategy. Software makers tailored their products to run on the ubiquitous Windows operating system. Then Apple’s iOS operating system for phones and tablets became a foundation for mobile developers to reach users. Over the years, companies like Intel, Cisco, IBM, and even AT&T built platforms and then reaped the rewards of that advantageous position.

So it was only natural that as early as 1997, executives at Amazon were thinking about how to become a platform and augment the e-commerce efforts of other retailers. Amazon Auctions was the first such attempt, followed by zShops, the service that allowed small retailers to set up their own stores on Amazon.com. Both efforts failed in the face of eBay’s insurmountable popularity with mom-and-pop merchants. Nevertheless, by 2000, according to an internal company memo, Bezos was telling colleagues that by the time Amazon got to $200 billion in annual sales, he wanted revenues to be split evenly between sales from products it sold itself and commissions that it collected from other sellers who used Amazon.com.

Ironically, it was the industrywide overreach of 1999 that finally sent Amazon down the path of becoming a platform. Toys “R” Us, though it had taken a $60 million investment from SoftBank and the private equity firm Mobius Equity Partners to create the Internet subsidiary ToysRUs.com, stumbled badly during the 1999 holidays. The offline retailer suffered a raft of negative publicity from frequent outages of its website and late shipments of orders, which in some cases missed Christmas altogether. The company ended up paying a $350,000 fine to the Federal Trade Commission for failing to fulfill its promises to customers. Amazon, meanwhile, had to write off $39 million in the unsold toy inventory that Bezos had so fervently vowed he would personally drive to the local dump.

One night after the holidays, ToysRUs.com chief financial officer Jon Foster cold-called Bezos in his office, and the Amazon CEO picked up the phone. Foster suggested joining forces; the online retailer could provide the critical infrastructure, and the offline retailer would bring the product expertise and relationships with suppliers like Hasbro. Bezos suggested the Toys “R” Us execs meet with Harrison Miller, the category manager of the toy business. The companies held a preliminary meeting in Seattle, but at that point Amazon saw little reason to collaborate with a key competitor.

The next spring, Miller and Amazon’s operations team studied the problems of stocking and shipping toys and concluded that achieving profitability in the category would require sales of nearly $1 billion. The biggest challenge was selecting and acquiring just the right selection of toys—precisely the kind of thing Toys “R” Us did well.

A few weeks later, Miller and Mark Britto, who ran Amazon’s business-development group, met with ToysRUs.com executives in a tiny conference room at Chicago’s O’Hare International Airport and began formal negotiations to combine their toy-selling efforts. “It was dawning on us how brutal it was to pick Barbies and Digimons, and it was dawning on them how expensive it would be to build a world-class e-commerce infrastructure,” Miller says.

It seemed like a perfect fit. Toys “R” Us was adept at choosing the right toys for each season and had the necessary clout with manufacturers to get favorable prices and sufficient supplies of the most popular toys. Amazon of course had the expertise to run an online retailing business and get products to customers on time. The negotiations were, as was often the case when Jeff Bezos was involved, long and, according to Jon Foster, “excruciating.” When both teams met for the first time, Bezos made a big show of keeping one chair open at the conference-room table, “for the customer,” he explained. Bezos was primarily focused on building comprehensive selection and wanted Toys “R” Us to commit to putting every available toy on the site. Toys “R” Us argued that this was impractical and expensive. Meanwhile, it wanted to be the exclusive seller of toys on Amazon.com, which Bezos felt was too constricting.

The companies were at loggerheads for months. To get the deal done, they met somewhere in the murky middle. Toys “R” Us agreed to sell the few hundred most popular toys, and Amazon reserved the right to complement the Toys “R” Us selection with less popular items. Neither company got what it wanted, but for the moment, everyone was relieved. In August, the companies announced a ten-year partnership, with Amazon getting a major source of desperately needed cash and some help with its balance-sheet problems. The companies agreed that Toys “R” Us inventory would be kept in Amazon’s distribution centers—the first step toward making the most expensive and complicated part of Amazon’s business a platform that other companies could use.

The deal became a template for Amazon. Having outsourced his job running the toy category to Toys “R” Us, Harrison Miller assumed a newly created role as head of platform services. With Neil Roseman, a vice president of engineering, he started traveling the country pitching other big retailers on duplicating the Toys “R” Us deal.

They came close with electronics giant Best Buy, until the chain’s founder, Richard Schulze, insisted late in the negotiations during a dramatic Saturday-morning conference call that Amazon give him total exclusivity in the electronics category. Bezos refused. Bed, Bath, and Beyond and Barnes & Noble also balked.

Sony Electronics explored the possibility of using Amazon to bring its Sony Style chain online. As part of the discussions, Howard Stringer, chief of Sony Corporation of America, toured the Amazon fulfillment center in Fernley and, in a memorable moment, encountered on the warehouse floor a pile of Sony merchandise, which Amazon was technically not supposed to be selling. Stringer and his colleagues started examining the labels and writing down product numbers in an attempt to determine where the merchandise had come from. That deal didn’t happen either.

But in early 2001, the effort started to gain traction. Amazon signed a deal with the book chain Borders, which had blundered by building a massive distribution facility outside Nashville for online orders before realizing it needed smaller, geographically dispersed warehouses to get books to customers quickly and inexpensively. A few months later, Amazon agreed to run AOL’s shopping channel in return for a much-needed $100 million investment. Amazon also signed a deal to carry the inventory of retailer Circuit City, helping to add additional selection to the sparsely furnished shelves of Amazon’s electronics category.

All of these deals improved Amazon’s balance sheet in the short term, but in the long run, they were awkward for all parties. By relying on Amazon, the retailers delayed a necessary education on an important new frontier and ceded the loyalty of their customers to an aggressive upstart. That would be one of many problems for Borders and Circuit City, both of which went bankrupt in the depths of the financial crisis that began in 2008.

Bezos never got completely comfortable with these deals or with the idea of outsourcing his prized goal of limitless selection. The Toys “R” Us arrangement in particular was hugely lucrative, but Bezos grew frustrated as it became more difficult to ensure that Amazon could offer a comprehensive toy selection. That ultimately factored heavily into the outcome of the partnership several years later—dueling lawsuits in federal court.

In the summer of 2000, with Ravi Suria continuing to press his case in public, the slide in Amazon’s stock price started to accelerate. In the span of three weeks in June, it dropped from $57 to $33, shedding almost half its value. Employees started to get nervous. Bezos scrawled I am not my stock price on the whiteboard in his office and instructed everyone to ignore the mounting pessimism. “You don’t feel thirty percent smarter when the stock goes up by thirty percent, so when the stock goes down you shouldn’t feel thirty percent dumber,” he said at an all-hands meeting. He quoted Benjamin Graham, the British-born investor who inspired Warren Buffett: “In the short term, the stock market is a voting machine. In the long run, it’s a weighing machine” that measures a company’s true value. If Amazon stayed focused on the customer, Bezos declared, the company would be fine.

As if to prove his singular obsession with customer experience, Bezos placed an expensive bet, hitching Amazon’s Quidditch broom to the rising fantasy series Harry Potter. In July, author J. K. Rowling published the fourth book in the series, Harry Potter and the Goblet of Fire. Amazon offered a 40 percent discount on the book and express delivery so customers would get it on Saturday, July 8—the day the book was released—for the cost of regular delivery. Amazon lost a few dollars on each of about 255,000 orders, just the kind of money-losing gambit that frustrated Wall Street. But Bezos refused to see it as anything other than a move to build customer loyalty. “That either- or mentality, that if you are doing something good for customers it must be bad for shareholders, is very amateurish,” he said in our interview that summer.

The Harry Potter promotion unsettled even the executives working on it. “I was thinking, Holy shit, this is a lot of money,” says Lyn Blake, the Amazon executive in charge of books at the time. She was later inclined to admit that Bezos was right. “We were able to assess all the good press and heard all these stories from people who were meeting their deliverymen at their front doors. And we got these testimonials back from drivers. It was the best day of their lives.” Amazon was mentioned in some seven hundred stories about the new Harry Potter novel in June and July that year.

Bezos was obsessed with the customer experience, and anyone who didn’t have the same single-minded focus or who he felt wasn’t demonstrating a capacity for thinking big bore the brunt of his considerable temper. One person who became a frequent target during this time was the vice president in charge of customer service, Bill Price.

A veteran of long-distance provider MCI, Price came to Amazon in 1999. He blundered early by suggesting in a meeting that Amazon executives who traveled frequently should be permitted to fly business-class. Bezos often said he wanted his colleagues to speak their minds, but at times it seemed he did not appreciate being personally challenged. “You would have thought I was trying to stop the Earth from tilting on its axis,” Price says, recalling that moment with horror years later. “Jeff slammed his hand on the table and said, ‘That is not how an owner thinks! That’s the dumbest idea I’ve ever heard.’

“Of course everyone else was thinking [executives should be allowed to fly business-class], but I was the exposed nail in the room,” Price says.

The 2000 holidays would be Price’s Waterloo. His customer-service department tracked two important metrics: average talk time (the amount of time an employee spent on the phone with a customer) and contacts per order (the number of times a purchase necessitated a customer phone call or e-mail). Bezos demanded that Price reduce both, but that was fundamentally impractical. If a customer-service rep stayed on the phone long enough to fully solve each customer’s problem, the number of contacts per order might go down, but the average talk time would go up. If the customer-service rep tried to jump off each call quickly, average talk times would decline, but customers would be more likely to call back.

Bezos didn’t care about that simple calculus. He hated when customers called at all, seeing it as a defect in the system, and he believed that customers should be able to solve their problems themselves with the aid of self-help tools.5 When they did call, Bezos wanted their queries answered promptly and their issues settled conclusively. There were no excuses. Price’s only solution was to push his team harder, but since he had limited resources to add new people, employees were burning out.

The denouement came in a new S Team ritual called the war room, a meeting that was held daily during the holiday period to review critical company and customer issues. About thirty senior executives in the company packed into a conference room on the top floor of the Pac Med building that had expansive views of the Puget Sound. With Christmas sales ramping up and hold times on Amazon phone lines once again growing longer, Bezos began the meeting by asking Price what the customer wait times were. Price then violated a cardinal rule at Amazon: he assured Bezos that they were well under a minute but without offering much in the way of proof.

“Really?” Bezos said. “Let’s see.” On the speakerphone in the middle of the conference table, he called Amazon’s 800 number. Incongruously cheerful hold music filled the room. Bezos took his watch off and made a deliberate show of tracking the time. A brutal minute passed, then two. Other execs fidgeted uncomfortably while Price furtively picked up his cell phone and quietly tried to summon his subordinates. Bezos’s face grew red; the vein in his forehead, a hurricane warning system, popped out and introduced itself to the room. Around four and a half minutes passed, but according to multiple people at the meeting who related the story, the wait seemed interminable.

Eventually a cheerful voice blurted out, “Hello, Amazon.com!” Bezos said, “I’m just calling to check,” and slammed down the phone. Then he tore into Price, accusing him of incompetence and lying.

Price was gone about ten months later.

While Amazon executives were courting large chain retailers, a rival was courting them. CEO of eBay Meg Whitman and one of her top deputies, Jeff Jordan, visited Amazon that fall with a tempting proposal: they wanted to take over Amazon’s failing Amazon Auctions business.

Whitman made a convincing case. She highlighted eBay’s focus on the unruly community of small sellers and argued that Amazon’s core retail business was at fundamental odds with its attempts to host third-party sellers, since both Amazon and these merchants were often competing to sell the same items. However, eBay had no such conflict, since it didn’t sell anything itself. Whitman argued that the deal could solve a problem for Amazon while also strengthening eBay’s position in its primary area of focus, auctions. It was the classic win-win scenario.

But Bezos declined the offer for the same reason he kept the ghost towns of Auctions and zShops alive on the Amazon website. He wasn’t ready to give up or relinquish Amazon’s hopes of becoming a platform for small and midsized retailers. The fact that third-party selling on Amazon wasn’t working meant, to Bezos, only that it wasn’t working at that particular moment.

The main problem was that Auctions and zShops were siloed on the Amazon website and got little attention from customers. Bezos referred to them as cul-de-sacs on the site. To the extent they enjoyed any traffic at all, it was through a feature called Crosslinks, in which links to third-party auctions appeared on related retail pages. For example, a seller hawking vintage fishing rods could choose to list his auctions via Crosslinks on the pages of books or movies about fly-fishing.6

Amazon experimented with using algorithms to analyze specific phrases on product pages and auctions and then automatically matching up similar products. The technology resulted in some memorable miscues. For example, the product page for a novel titled The Subtle Knife, the sequel to the young-adult novel The Golden Compass, carried links to a variety of survivalist-minded sellers in the auctions category who were hawking switchblades and SS weaponry kits. “There were some very unhappy results,” says Joel Spiegel. “The person whose mission in life was to sell children’s books would storm into my office yelling, Why the hell do I have Nazi memorabilia listed on my pages?”

One weekend in the fall of 2000, Bezos called various S Team members and executives to a daylong meeting in the basement of his lakefront mansion in Medina so they could examine why the third-party efforts were failing. Despite the problems, the group recognized that Crosslinks on the product pages were generating most of the traffic to Amazon’s third-party sellers.

That was an important observation. Traffic on Amazon was oriented around Amazon’s reliable product catalog. On eBay, a customer might search for the Hemingway novel The Sun Also Rises and get dozens of auctions of new and vintage copies. If a customer searched for the book on Amazon, there was one single page, with a definitive description of the novel, and that’s where customers flocked.

Amazon executives reasoned that day that they had the Internet’s most authoritative product catalog and that they should exploit it. That, it turned out, was the central insight that not only turned Amazon into a thriving platform for small online merchants but powers a good deal of its success today. If Amazon wanted to host other sellers on its site, it would have to list their wares right alongside its own products on the pages that customers actually visited. “It was a great meeting,” says Jeff Blackburn. “By the end of the day we all felt one hundred percent sure that this was the future.”

That fall, Amazon announced a new initiative called Marketplace. The effort started with used books. Other sellers of books were invited to advertise their wares directly within a box on Amazon’s own book pages. Customers got to choose whether to purchase the item from Amazon itself or from a third-party seller. If they chose the latter, either because the seller had a lower price or because the product was out of stock at Amazon, the company would lose the sale but collect a small commission. “Jeff was super clear from the beginning,” says Neil Roseman. “If somebody else can sell it cheaper than us, we should let them and figure out how they are able to do it.”

Marketplace launched in November 2000 in the books category and immediately drew protests. Two trade groups, the Association of American Publishers and the Authors Guild, each posted a public letter on its website complaining that Amazon was undermining the sale of new books in favor of used books and in the process taking royalties out of the pockets of authors.7 “If your aggressive promotion of used book sales becomes popular among Amazon’s customers, this service will cut significantly into sales of new titles, directly harming authors and publishers,” said the letter.

The protest was nothing compared to the consternation over Marketplace inside Amazon. Category managers realized they could now lose a sale to a competitor within the previously safe confines of their own store. Even worse, a customer might have a bad experience with that seller and end up leaving a negative review. And the company’s buyers now had to contend with irate publishers and other manufacturers who wanted to know why used products from small, often unauthorized sellers were being sold directly next to their new wares. This debate would play out gradually over the next few years as Amazon expanded the effort and added both new and used products from third-party sellers to each category. Marketplace, in effect, made it more difficult for the retailers inside Amazon to accomplish the lofty goals Bezos himself had set for them.

“Imagine you’re the guy on the hook for a zillion dollars’ worth of inventory,” says Chris Payne, recalling his initial reaction to Marketplace. “And this other lunatic comes over putting low-priced crap on your page. You can bet that leads to some squabbles.”

The new strategy would result in years of tension between various divisions, between Amazon and its suppliers, and between industry trade groups and the company. Bezos didn’t care about any of that, as long as it offered more choices to customers and, in the process, gave Amazon a greater selection of products. With a single brilliant and nonintuitive strategic move, he managed to upset almost everybody, even his own colleagues. “As usual,” says Mark Britto, “it was Jeff against the world.”

One Saturday in early December 2000, Britto and Doug Boake, business-development executives who joined Amazon in the Accept.com acquisition, were in Fernley gift-wrapping packages when Britto got a call on his cell phone. It was Bezos. He told them to meet him that night in Bentonville, Arkansas. They were going to visit Walmart.

Though it sounds unlikely, now that they are archrivals, Amazon was pitching Walmart on the idea of operating its website. Walmart was the undisputed gorilla of retailing, opening hundreds of new stores a year around the world and remaining relatively unharmed by the bear market. Lee Scott, just the third CEO in Walmart history, had personally invited Bezos to his home. Britto and Boake happily put down the gift wrap and headed to the Reno airport.

That evening, the Amazon executives met in Bentonville, where they got a taste of Walmart’s brand of frugality. Walmart booked them rooms at a local Days Inn. That night, Bezos, Britto, and Boake had dinner at a nearby Chili’s and walked around the historic town square.

The next morning, a procession of three black Chevy Suburbans rolled up to the hotel at the appointed time. The drivers wore earpieces, sunglasses, and steely expressions. The Amazon executives were ushered into the middle car and marveled at the abundance of security. Though he didn’t know it, Bezos was glimpsing his own future.

The cars drove to a large house in a gated community off a golf course, and the Amazon execs got out, walked up, and knocked on the front door. Linda Scott, the CEO’s wife, opened the door and immediately put them at ease. She told Bezos she was a big fan of his and had watched his appearance on CNBC’s Squawk Box a few weeks before.

The Amazon execs met Lee Scott and his chief financial officer, Tom Schoewe, in a dining room with big bay windows. For two hours, over pastries and coffee, the CEOs spoke frankly. They talked about the companies’ shared culture and the principles Bezos had taken from Sam Walton’s autobiography. Bezos spoke generally about Amazon’s attempts at personalization and the technology behind collaborative filtering—the algorithms that determined that people who bought one particular kind of product were inclined to purchase another specific set of products.

Scott noted that Walmart had similar techniques. It could measure whether a certain item, such as a globe for children, could lift the sale of another item, like a coloring book, if they were placed next to each other on a store display. Both companies had a deep interest in testing these combinations.

Scott also talked about how Walmart viewed advertising and pricing as two ends on the same spectrum. “We spend only forty basis points on marketing. Go look at our shareholder statement,” he said. “Most of that goes to newspapers to inform people about what is in our stores. The rest of our marketing dollars we pour into reducing prices. Our marketing strategy is our pricing strategy, which is everyday low pricing.”

Before the meeting, Rick Dalzell had warned Bezos to be wary of the crafty and astute Walmart chief. But Bezos was sponging up everything the older man said. Amazon had always considered itself an e-commerce company, not a retailer. Now Bezos needed to learn some of the fundamental rules in a professional sport that, up until that point, he had been playing only amateurishly.

After the first hour, the executives got down to business. Scott wanted to know what Amazon had in mind. The execs explained the Toys “R” Us deal and the nascent effort to operate the websites and handle distribution for other retailers. Scott said noncommittally that it was worth talking about. To conclude the meeting, he leaned forward and said, “So, is there something deeper and more strategic that we should be considering?”

Bezos said he would think about how to make the proposal more interesting to Walmart. The men shook hands, and the Amazon executives returned to the Suburban waiting out front. As they were being driven to the airport, Britto and Boake agreed that Lee Scott’s parting words could be interpreted only as a veiled acquisition offer. “Really, is that what he meant by that?” Bezos asked.

Of course Bezos wasn’t interested in selling his company to Walmart, and Scott ultimately rejected the idea of outsourcing a crucial part of Walmart’s online operation to Amazon. The conversation between the two retailers never developed further and the meeting remained a quirk of history, a tantalizing suggestion of what might have been. The two companies would continue on separate paths, which, years later, would converge to produce a fierce rivalry.

In February 2001, Ravi Suria reared his head again. He published another report that questioned Amazon’s reserve of capital. With Amazon facing $130 million in annual interest expenses on its debt and given the prospect of its continued losses, Suria predicted that the company would face a cash shortage by the end of the year.

This time, Amazon made it personal. Spokesman Bill Curry retorted in an interview that Suria’s report was “silly.”8 Warren Jenson paid a personal visit to Lehman vice chairman Howard Clark, and John Doerr called Dick Fuld, chief executive of the investment bank, and implored him to have the firm review Suria’s research.

Years later, over cocktails at midtown Manhattan’s Trump Bar, with its dim lights and dark, polished wood, Suria complained that Amazon exerted unbearable pressure on him during that time. “They wanted to fire me. Everyone at Lehman hated my guts during those months,” he says. “Every time I picked up the phone someone was screaming at me.”

Suria now helps to run a hedge fund and has a bitter view of his history with the online retailer. “Amazon was like a high-school bully picking on an elementary-school kid. I was twenty-nine years old. It was a character-defining moment [for them], and as far as I’m concerned, they failed it miserably. It ruined my life for two years.” Suria believes Bezos is “deranged” and proudly notes that he hasn’t bought anything from Amazon since he tangled with the company.

But there’s no doubt that investors were keyed into Suria’s analysis. The February research report, his last at Lehman Brothers before departing for the hedge fund Duquesne Capital Management, sent Amazon stock roaring toward the ignominious land of the single digits. It had another repercussion as well. In regulatory papers filed by his lawyer that month, Bezos revealed intentions to sell a small parcel of stock, worth about $12 million. Since Lehman had allowed Amazon to see a version of Suria’s report before it was published, the timing of the stock sale suggested to the Securities and Exchange Commission that Bezos was deliberately dumping Amazon shares before bad news was made public.

In retrospect, one can see it was the farthest thing from the truth; Bezos remained completely convinced of the eventual success of his venture. But the SEC—which had been hammered by critics for whiffing on the dot-com bubble—announced an investigation into the possibility of insider trading. The investigation went nowhere, but the New York Times, among other publications, splashed the news prominently on the front of its business section.9 “I don’t care who you are or how much chutzpah you have,” says Warren Jenson. “It’s not fun picking up the Times and seeing your picture above the fold accused of insider trading. We are all products of what we’ve been through. This is one of the things that made Jeff the person he is. That scar does not heal easily.”

Now Amazon once again had to come to terms with the practical effects of its deteriorating stock price and its overzealous expansion. That month Amazon repriced the stock options of employees. They could trade three shares at their old stock price for one share at the new price—a move that boosted the morale of employees whose options, with the cratering stock prices, had become worthless. Amazon also announced plans to cut thirteen hundred employees, or about 15 percent of its workforce. The company was accustomed to adding people, not losing them, and the layoffs were brutal. People who had been hired just months before were summarily fired, their careers and personal lives left in tatters. Mitch Berman, a merchandising manager in the DVD group, had previously worked at Coca-Cola in Atlanta and had moved to Seattle for the job. He was employed by Amazon for all of four months and never understood why it didn’t work out. “I had literally picked up my entire life and moved across the country,” he says. “Obviously, I felt burned. I had to roll up my sleeves and start all over again.” He’s now a life coach living in Barcelona, Spain.

Diego Piacentini, a new executive from Apple, was thrust directly into the mess. Bezos hired the suave, Italian-born Piacentini in early 2000 to take the top spot running Amazon’s international operations. Piacentini’s old boss Steve Jobs had expressed incredulity at the move in his typically strident way. Over lunch in the Apple cafeteria in Cupertino, Jobs asked Piacentini why he would possibly want to go to a boring retailer when Apple was in the process of reinventing computing. Then in the same breath, Jobs suggested that maybe the career move revealed that Piacentini was so dumb that it was a good thing he was leaving Apple.

At first, Piacentini himself wondered why he’d made the move. He had joined Amazon right in the middle of Bezos’s conflict with Joe Galli. After his first few weeks, Piacentini called his wife back in Milan and told her not to pack their things for Seattle quite yet. But after Galli left, he grew more comfortable at Amazon. A year later, during the layoffs, he was tasked with closing Amazon’s new multilingual call center in The Hague. The facility had been poorly selected. The Hague was a financial and diplomatic hub, and the call center was incongruously located in a marble-floored building that had once been occupied by a bank. It never should have been opened in the first place, but “people at various levels were making decentralized decisions to move quickly and the process wasn’t strong,” Piacentini says.

The center had been open only a few months when Piacentini arrived to shut it down. With a few colleagues from Seattle, he collected the two hundred and fifty or so employees in the large marble lobby and made a brief speech in English telling everyone the bad news. Employees started howling and shouting, according to one Amazon employee who was there. One woman began sobbing and rolling on the floor.

Inside Amazon’s Seattle offices, it seemed like the walls were closing in—at times, literally. On the morning of Wednesday, February 28, Neil Roseman, Rick Dalzell, and an executive named Tom Killalea met with Bezos in his private conference room to brief him on a potentially serious security breach at Amazon’s used-book marketplace, Exchange.com. A few minutes into the conversation, the room started to shake.

It started slowly, a rumbling in the floor that passed into the walls and then intensified. The four men looked questioningly at one another and then dove under the side-by-side door-desks at the center of the room. Forty-six miles southwest, the Earth had suddenly shifted, and the Nisqually earthquake, 6.9 on the Richter scale, had begun.

Outside, chunks of brick and mortar were shaken loose from the sixty-eight-year-old Pacific Medical building and rained to the ground. Inside, the sprinklers went off and employees rolled under their mercifully thick door-desks. Bezos’s tiny conference room was full of tchotchkes like Star Trek figurines and water guns, many of which noisily rattled to the floor. Also in the room was a twenty-two-pound ball made of the dense metal tungsten, a memento from Stewart Brand and the organizers of the Clock of the Long Now. Halfway through the earthquake, the executives in the room heard the ominous sound of the ball rolling off its stand. “I was the low man on the totem pole, so my legs were halfway exposed,” says Neil Roseman, only partly in jest. Fortunately, the ball thudded harmlessly to the floor.

As the earthquake progressed, Killalea poked his head out, retrieved his laptop, and checked to see if the Amazon website was still running. (He would win a Just Do It award and get to keep an old ratty sneaker for that bit of bravado.)

The rumbling stopped after forty-five seconds, and employees evacuated the building. In a commanding performance, Bezos donned an item from his collection of oddities, a hard hat shaped like a ten-gallon cowboy hat, scrambled onto the roof of a car in the parking lot, and organized pairs of employees to reenter the building and collect their valuables. The building owner later shut down the tenth and twelfth floors for repairs, and for months plastic tarps covered patches of the façade where bricks had shaken loose.

When I visited Amazon for another Newsweek story that March, the stock was hovering around $10 and the city inspector had closed the main lobby. It was an unattractive sight and an unavoidable metaphor for the company’s rapid descent. Visitors were ushered into the basement through the back of the building, past a large placard warning of falling bricks.

In early 2001, Amazon’s position and future prospects remained dubious. The problem wasn’t only its diminishing market capitalization or its overlarded staffing and expansion efforts. Growth, particularly in the oldest category, books—still more than half its business at the time—appeared to be slowing after years of annual double-digit increases. Inside the company, executives were fearful that the slowdown augured an overall decrease in online shopping itself. “We were scared to death,” says Erich Ringewald, a vice president in charge of Marketplace. “Books were decelerating, and everyone thought that Walmart.com would start selling books at a loss to keep us from growing.”

Amazon then did something rare in its history. Warren Jenson, pushing to improve margins to meet the company’s self-imposed profitability deadline, convinced Bezos to quietly raise prices in the older media categories. Amazon reduced its discounts on bestselling books and started charging more to overseas customers who were buying from the domestic website. Bezos signed off on the increases, but another important meeting quickly made him change his mind.

On a Saturday morning that spring, at the Starbucks inside the Bellevue Barnes & Noble where he had conducted Amazon’s very first meetings, Bezos met Jim Sinegal, the founder of Costco. Sinegal was a casual, plain-speaking native of Pittsburgh, a Wilford Brimley look-alike with a bushy white mustache and an amiable countenance that concealed the steely determination of an entrepreneur. Well into retirement age, he showed no interest in slowing down. The two had plenty in common. For years Sinegal, like Bezos, had battled Wall Street analysts who wanted him to raise Costco’s prices on clothing, appliances, and packaged foods. Like Bezos, Sinegal had rejected multiple acquisition offers over the years, including one from Sam Walton, and he liked to say he didn’t have an exit strategy—he was building a company for the long term.

Bezos had set up the meeting to ask Sinegal about using Costco as a wholesale supplier for products that manufacturers still wouldn’t sell to Amazon. That idea never went anywhere, but over the next hour, Bezos listened carefully and once again drew key lessons from a more experienced retail veteran.

Sinegal explained the Costco model to Bezos: it was all about customer loyalty. There are some four thousand products in the average Costco warehouse, including limited-quantity seasonal or trendy products called treasure-hunt items that are spread out around the building. Though the selection of products in individual categories is limited, there are copious quantities of everything there—and it is all dirt cheap. Costco buys in bulk and marks up everything at a standard, across-the-board 14 percent, even when it could charge more. It doesn’t advertise at all, and earns most of its gross profit from the annual membership fees.

“The membership fee is a onetime pain, but it’s reinforced every time customers walk in and see forty-seven-inch televisions that are two hundred dollars less than anyplace else,” Sinegal said. “It reinforces the value of the concept. Customers know they will find really cheap stuff at Costco.”10

Costco’s low prices generated heavy sales volume, and the company then used its significant size to demand the best possible deals from suppliers and raise its per-unit gross profit dollars. Its vendors hadn’t been happy about being squeezed but they eventually came around. “You can fill Safeco Field with the people that don’t want to sell to us,” Sinegal said. “But over a period of time, we generate enough business and prove we are a good customer and pay our bills and keep our promises. Then they say, ‘Why the hell am I not doing business with these guys. I gotta be stupid. They are a great form of distribution.’

“My approach has always been that value trumps everything,” Sinegal continued. “The reason people are prepared to come to our strange places to shop is that we have value. We deliver on that value constantly. There are no annuities in this business.”

A decade later and finally preparing to retire, Sinegal remembers that conversation well. “I think Jeff looked at it and thought that was something that would apply to his business as well,” he says. Sinegal doesn’t regret educating an entrepreneur who would evolve into a ferocious competitor. “I’ve always had the opinion that we have shamelessly stolen any good ideas,” he says.

In 2008, Sinegal bought a Kindle e-reader that turned out to be defective and wrote Bezos a laudatory e-mail after Amazon’s customer service replaced his device for free. Bezos wrote back, “I want you to consider me your personal customer service agent on the Kindle.”

Perhaps Amazon’s founder realized he owed Sinegal a debt of gratitude, because he took the lessons he learned during that coffee in 2001 and applied them with a vengeance.

The Monday after the meeting with Sinegal, Bezos opened an S Team meeting by saying he was determined to make a change. The company’s pricing strategy, he said, according to several executives who were there, was incoherent. Amazon preached low prices but in some cases its prices were higher than competitors’. Like Walmart and Costco, Bezos said, Amazon should have “everyday low prices.” The company should look at other large retailers and match their lowest prices, all the time. If Amazon could stay competitive on price, it could win the day on unlimited selection and on the convenience afforded to customers who didn’t have to get in the car to go to a store and wait in line.

That July, as a result of the Sinegal meeting, Amazon announced it was cutting prices of books, music, and videos by 20 to 30 percent. “There are two kinds of retailers: there are those folks who work to figure how to charge more, and there are companies that work to figure how to charge less, and we are going to be the second, full-stop,” he said in that month’s quarterly conference call with analysts, coining a new Jeffism to be repeated over and over ad nauseam for years.

Bezos had seemingly made up his mind that he was no longer going to indulge in financial maneuvering as a way to escape the rather large hole Amazon had dug for itself, and it wasn’t just through borrowing Sinegal’s business plan. At a two-day management and board offsite later that year, Amazon invited business thinker Jim Collins to present the findings from his soon-to-be-published book Good to Great. Collins had studied the company and led a series of intense discussions at the offsite. “You’ve got to decide what you’re great at,” he told the Amazon executives.

Drawing on Collins’s concept of a flywheel, or self-reinforcing loop, Bezos and his lieutenants sketched their own virtuous cycle, which they believed powered their business. It went something like this: Lower prices led to more customer visits. More customers increased the volume of sales and attracted more commission-paying third-party sellers to the site. That allowed Amazon to get more out of fixed costs like the fulfillment centers and the servers needed to run the website. This greater efficiency then enabled it to lower prices further. Feed any part of this flywheel, they reasoned, and it should accelerate the loop. Amazon executives were elated; according to several members of the S Team at the time, they felt that, after five years, they finally understood their own business. But when Warren Jenson asked Bezos if he should put the flywheel in his presentations to analysts, Bezos asked him not to. For now, he considered it the secret sauce.

In September 2001, Bezos, Mark Britto, Harrison Miller, and two Amazon publicists flew to Minneapolis to announce a long-planned deal with Target. On the day of the announcement, they arrived just before 8:00 a.m. at the retailer’s downtown headquarters and took an elevator to a television studio on the thirty-second floor of Target Plaza South, the tallest building in the city. As they were in the elevator, Amazon PR chief Bill Curry got a call from a colleague in Seattle. A plane had hit the World Trade Center. When they got upstairs, they asked their Target counterparts to turn on the television.

Together the Amazon and Target executives watched in horror as the second plane hit the World Trade Center. No one had any idea what was going on. Curry, a former publicist for Boeing, observed that the plane looked like a 767. Plans to publicize the partnership with a series of satellite-television interviews were scrubbed. The tragic morning then unfolded before them, as it did for everyone else around the world. The Target building was evacuated and then reopened, and the Amazon and Target executives stood together for much of the day watching a single television.

In the afternoon, Bezos, still on his photography kick, walked around the Target office taking pictures with his Elph digital camera to record the awful, historic day. Someone complained to Dale Nitschke, the Target manager in charge of the Amazon partnership, and he quietly asked Bezos to stop.

The skies were closed to commercial flights for the next seventy-two hours, so the Amazon group couldn’t fly back to Seattle. On the morning of September 12, they bought additional clothes and an automobile cell phone charger from a Marshall Field’s department store, rented a white Mazda minivan from Hertz at an exorbitant daily rate, and headed west on I-90, a highway that ended in Seattle.11 Britto drove, Miller sat in the front seat, and they all stewed, shell-shocked, listening to music and their own thoughts. “Driving through the farmland and thinking about what was next was surreal and cathartic,” Miller says.

While Britto drove, Bezos used his phone and helped to organize a donation drive on the Amazon home page, which in two weeks would raise seven million dollars for the Red Cross. They stopped to stretch their legs in the Badlands and spent the night at a Mount Rushmore hotel that Bezos remembered visiting with his family as a child. Flags were at half-mast at the Mount Rushmore memorial, and the tourists were somber. Some tourists recognized Bezos—not as the Amazon.com founder, but as the CEO who had just appeared in a goofy ad for Taco Bell to raise money for the Special Olympics. Afterward, the executives bought matching navy blue Mount Rushmore windbreakers and ate at the park cafeteria.

The group kept driving west. Later that day, the skies briefly reopened for private flights, and Bezos’s plane met them on a small airstrip. Bowing to the gravity of the moment, Bezos did not make his usual announcement that the company was not paying for the flight; they flew to Seattle, and their solemn cross-country odyssey ended.

Bezos may have been famous to some because of his notorious Taco Bell ad, but in fact Amazon had some of the most memorable TV ads of the dot-com era. In the Sweatermen series, created by the San Francisco office of an agency called FCB Worldwide, a campy chorus of men dressed like Mr. Rogers extolled the virtues of unlimited selection on Amazon. The playful, retro shtick reflected the goofy sensibility of Amazon’s CEO. But a year into the bust, Bezos was desperately trying to figure out how he could stop advertising altogether.

As usual, Bezos battled his marketing executives. They argued that Amazon had to be on the airwaves to reach new customers. As Amazon’s losses mounted, Bezos’s opposition hardened. He had the marketing department organize tests, running commercials in only the Minneapolis and Portland media markets and measuring whether they generated an uptick in local purchases. They did—but, Bezos concluded, not enough to justify the investment.12 “It was pretty clear afterward that TV advertising wasn’t really having an impact,” says Mark Stabingas, a finance vice president who joined the company from Pepsi.

The result was not only the cancellation of all of Amazon’s television advertising but another dramatic purge of the marketing department. Alan Brown, a chief marketing officer who came from MasterCard, left after only a year on the job. Centralized marketing at Amazon was shut down and its tasks spread out among the e-mail marketing and worldwide discovery groups led by Andy Jassy and Jeff Holden. Amazon wouldn’t advertise on television again for another seven years, not until the introduction of the Kindle. “There can be only one head of marketing at Amazon, and his name is Jeff,” says Diane Lye, a British senior manager who led Amazon’s data-mining department and helped run the advertising tests.

Bezos felt that word of mouth could deliver customers to Amazon. He wanted to funnel the saved marketing dollars into improving the customer experience and accelerating the flywheel. And as it happened, at the time, Amazon was conducting an experiment that was actually working this way—free shipping.

During the 2000 and 2001 holidays, Amazon offered free shipping to customers who placed orders of a hundred dollars or more. The promotion was expensive but clearly boosted sales. Customer surveys showed that shipping costs were one of the biggest hurdles to ordering online. Amazon hadn’t yet found a good way to convince customers to shop in multiple categories—to buy books, kitchen appliances, and software, for example, all at the same time. The hundred-dollar threshold motivated buyers to fill their baskets with a variety of items.

In early 2002 late on a Monday night, Bezos called a meeting in Warren Jenson’s conference room to talk about how to turn the holiday-season free shipping into a permanent offer. This was one way he could redeploy his marketing budget. Jenson in particular was opposed to this. The CFO worried that free shipping would be expensive and wasteful, since Amazon would be giving discounts to all comers, including those customers who were inclined to place large orders anyway.

Then one of his deputies, a finance vice president named Greg Greeley, mentioned how airlines had segmented their customers into two groups—business people and recreational travelers—by reducing ticket prices for those customers who were willing to stay at their destination through a Saturday night. Greeley suggested doing the equivalent at Amazon. They would make the free-shipping offer permanent, but only for customers who were willing to wait a few extra days for their order. Just like the airlines, Amazon would, in effect, divide its customers into two groups: those whose needs were time sensitive, and everyone else. The company could then reduce the expense of free shipping, because workers in the fulfillment centers could pack those free-shipping orders in the trucks that Amazon sent off to express shippers and the post office whenever the trucks had excess room. Bezos loved it. “That is exactly what we are going to do,” he said.

Amazon introduced the service, called Free Super Saver Shipping, in January 2002 for orders above $99. In the span of a few months, that number dropped to $49, and then to $25. Super Saver Shipping would set the stage for a variety of new initiatives in the years ahead, including the subscription club Amazon Prime.

Not everyone was happy with this outcome. After that meeting, Warren Jenson took Greeley aside and berated him, in that moment seeing free shipping as nothing but another potential balance-sheet buster.

Over the next year, Amazon executives quit in droves. They left because their stock had been vested or because they no longer believed in the mission or because their comparatively low salaries and the depressed stock price guaranteed that they were not getting wealthy anytime soon. Some were tired and just wanted a change. Others felt Bezos didn’t listen to them and that he wasn’t about to start. Almost all figured that Amazon’s best days were behind it. The company reached incredible levels of attrition in 2002 and 2003. “The number of employees at that point other than Jeff who thought he could turn it into an eighty-billion-dollar company—that’s a short list,” says Doug Boake, who departed for the Silicon Valley startup OpenTable. “He just never stopped believing. He never blinked once.”

They all had their reasons. David Risher left to teach at the University of Washington’s business school. Joel Spiegel wanted to spend more time with his three teenage kids before they left home. Mark Britto wanted to get back to the Bay Area. Harrison Miller was exhausted and needed a change. Chris Payne left for Microsoft, where he would help launch the Bing search engine, after which he would end up as a top executive at eBay. And on and on.

People left and afterward they took a breath and felt disoriented, like they had escaped a cult. Though they didn’t share it openly, many just couldn’t take working for Bezos any longer. He demanded more than they could possibly deliver and was extremely stingy with praise. At the same time, many felt a tremendous loyalty to Bezos and would later marvel at how much they accomplished at Amazon. Kim Rachmeler shared a favorite quote she heard from a colleague around that time. “If you’re not good, Jeff will chew you up and spit you out. And if you’re good, he will jump on your back and ride you into the ground.”

Bezos never despaired over the mass exodus. One of his gifts, his colleagues said, was being able to drive and motivate his employees without getting overly attached to them personally. But he did usually make time in his calendar for a private meeting with exiting executives. Harrison Miller told Bezos at their parting lunch that the accomplishment at Amazon he was most proud of was the platform-services business with large retailers, which was responsible for a third of Amazon’s cash flow in 2002. “Yeah, but don’t forget, you built our first toy store and it was great,” Bezos said, another indication he remained more focused on his long-term goal of unlimited selection than on his short-term revenue-boosting partnerships, however lucrative.

At his going-away meeting, Brian Birtwistle wrote a list on a cocktail napkin of his favorite moments at Amazon. Bezos and Birtwistle took a picture with the napkin and recalled their drive from Harvard’s business school to the Boston airport. “This whole journey started with a car ride and what a ride it’s been,” Bezos said.

Bezos wasn’t always so sentimental. Christopher Zyda, Amazon’s treasurer, defected to eBay, and, in a throwback to Walmart’s lawsuit over poaching, Amazon sued eBay in federal court, alleging that Zyda was violating the noncompete clause in his employment contract. The lawsuit, like the Walmart poaching case against Amazon, was settled with no consequential damages. But if his legal strategy was any guide, Bezos was clearly worried about high-flying eBay, whose market capitalization now exceeded Amazon’s by a significant margin.

The elevated competition between the two companies put at least one person in an awkward position: Scott Cook, Intuit’s founder, was still on the boards of both companies. Now it was clear he would need to cut ties with one of them. He chose to leave Amazon and stick with eBay. “Jeff was angry, but not at me,” Cook says. “He was angry at himself for not stopping me when I said I wanted to join the eBay board in the first place. He doesn’t like losing.”

Warren Jenson also left. Amazon’s CFO explained that he wanted to return to his wife and children, who were still living in Atlanta, and that the time was right because Amazon was finally clear of its most serious financial challenges. This was almost certainly not the whole story.

Jenson and Bezos were at loggerheads. Jenson had tried to placate angry investors by getting the company to profitability. He raised the last round of capital from European bonds at a critical time and forced Bezos to make tough decisions when the company’s runway was getting short. But he also pushed to raise prices and campaigned against free shipping. “I would never claim to be perfect,” he says. “I always tried to do what was right for the business.”

Jenson’s legacy at Amazon was hotly debated even a decade later. Some thought he was overly political. Others argued that he helped to direct the company away from its path of reckless growth and that he assembled an accomplished finance team that would go on to make significant contributions at Amazon and throughout the technology world. The evidence for the pro-Jenson case is difficult to dismiss. “Warren was the right CFO for the time,” says Dave Stephenson, a finance exec who worked for him at Amazon. “He forced hard decisions and hard debates. He would always stand up to Jeff a little bit more directly than anyone else.”

To replace Jenson, Bezos recruited another chief financial officer from General Electric, Tom Szkutak, sealing the deal with an impassioned two-page letter to Szkutak and his wife about the impact they could make at a historic juncture for the Internet. Szkutak was also the right CFO for Amazon at the right time. He would facilitate rather than challenge Bezos’s ambitious forays into various new businesses in the years ahead.

Perhaps the most rancorous exits in the company during this time stemmed from the intramural combat between two departments: Amazon’s editorial group and its personalization team. The editorial division, which dated back to Amazon’s earliest days, was composed of writers and editors who added a human touch to the Amazon home page and to the individual product pages. Bezos originally formed the group to cultivate the literary aura of an independent bookstore and recommend books to customers that they might not have otherwise found.

But over the years, the personalization group started to infringe on the editorial group’s turf. P13N, as it was cleverly abbreviated (there are thirteen letters between the p and the last n in personalization), used analytics and algorithms to generate recommendations crafted to appeal to individual customers based on their previous purchases. Over the years, P13N kept getting better. In 2001, Amazon started making suggestions based on the items customers looked at, not just the products they bought.

The juxtaposition between the two approaches was stark. Editorial was handselling products with clever writing and intuitive decisions about what to promote. (“We ain’t lion: this adorable Goliath Backpack Pal is a grrreat way to scare away those first-day-of-school jitters,” read the home page in 1999, promoting a lion-shaped backpack for kids.) Personalization was skipping the puns and building a store for every customer using cold, hard data to stock the shelves with the items that customers were statistically the most likely to buy.

Bezos did not explicitly favor one group over the other, but he looked at the results of tests. Over time it became clear that the humans couldn’t compete. PEOPLE FORGET THAT JOHN HENRY DIED IN THE END, read a sign on the wall of the P13N office, a reference to the folktale of the steel driver who raced to dig a hole in competition with a steam-powered drilling machine; he won the contest but died immediately afterward.

Most editors and writers were reassigned or laid off. Susan Benson—Rufus’s owner—took a sabbatical from Amazon. When she returned, Jason Kilar, then the vice president of media, invited her to a meeting that he described ominously in e-mails as an “editorial game changer.” She knew she was in trouble. “It had a lot do with how to dismantle editorial and turn it into part of the automated universe,” Benson says. “I thought, Yeah, my time here is done.”

An algorithm called Amabot brought about the downfall of editorial. Amabot replaced the personable, handcrafted sections of the site with automatically generated recommendations in a standardized layout. The system handily won a series of tests and demonstrated it could sell as many products as the human editors. Soon after, an anonymous Amazon employee placed a three-line classified advertisement in the Valentine’s Day 2002 edition of the Stranger, an independent Seattle newspaper. It read:

DEAREST AMABOT

If you only had a heart to absorb our hatred… Thanks for nothing, you jury-rigged rust bucket. The gorgeous messiness of flesh and blood will prevail!

In January 2002, Amazon reported its first profitable quarter, posting net income of $5 million, a meager but symbolic penny per share. Marketing costs were down, international revenues from the United Kingdom and Germany were up, and sales from third-party sellers on the vaunted Amazon platform made up 15 percent of the company’s orders. The exclamation point on the accomplishment was that Amazon had turned a profit by both controversial pro forma accounting standards and conventional methods.

Amazon had finally shown the world that it wasn’t just another doomed dot-com. The stock price immediately jumped 25 percent in after-hours trading, clawing its way out of the single digits. Kathy Savitt, a new Amazon publicist, told Bezos she wanted to frame some of the positive news articles and hang them on the office walls. He told her he would rather frame the negative stories like Barron’s infamous Amazon.bomb cover. When people wrote or said positive things about Amazon, he wanted employees to remember the Barron’s article and remain scared.

The company wasn’t yet entirely clear of its balance-sheet problems but it was on its way. In the first quarter the following year, Amazon cleared $1 billion in sales for the first time during a non-holiday period, setting the stage for its first profitable year. In a sign of optimism, Amazon said it would prematurely redeem the bonds from its first debt round back in 1998, paying bond holders the full outstanding value of the bonds five years before their maturity date.

As they prepared to make this announcement, someone on the finance team wondered what their old foe Ravi Suria was thinking. That revived the notion of the milliravi, a significant mathematical error. Mark Peek, the chief accounting officer at the time, joked that they should find a way to use the word in their press release. Everyone loved that idea, including Bezos, and they started exchanging suggestions over e-mail. Finally, investor-relations chief Tim Stone asked Bezos if he was serious about actually doing this, and Bezos said that yes, he definitely was.

Thus, on April 24, 2003, in the press release announcing quarterly earnings, shareholders, analysts, and journalists were treated to this inexplicable headline, which doubled as a quotation attributed to Bezos: MEANINGFUL INNOVATION LEADS, LAUNCHES, INSPIRES RELENTLESS AMAZON VISITOR IMPROVEMENTS.

Taking the first letter of each word and putting them together produced milliravi. A few of the analysts and reporters following the company scratched their heads over the unartful prose. No one outside Amazon knew what to make of it. But for Jeff Bezos, and for the employees who stuck with their implacably demanding leader through that first critical battle, the message was clear.

They had won.