One Click: Jeff Bezos and the Rise of Amazon.com Part 2

Chapter 5.

Three Nerds and an Accountant.

The Internet is like alcohol in some sense. It accentuates what you would do anyway. If you want to be a loner, you can be more alone. If you want to connect, it makes it easier to connect.

-Esther Dyson.

The first step in building a company is to find good people. And that can be a problem. How does one convince people to join a venture with almost no funding, and aspirations to create the world's biggest bookstore without actually creating a physical store and without actually stocking any books? Bezos's original idea was that he could simply order a book from the distributors once he had a buyer, and turn around and ship it off to the customer as soon as it arrived at his warehouse.

The process begins by tapping your business connections. It's often a matter of moving from one person to another, and from that person to yet another. It's a process encapsulated today in online networks such as the LinkedIn social networking site, where your connections can introduce you to other connections, who can introduce you to still more connections. Bezos didn't have LinkedIn, but he had made a lot of business connections, and he was good at exploiting them by phone. Bezos was prohibited from hiring people from D. E. Shaw, but he could use his comrades there as a starting point. MacKenzie wasn't the only happy connection Bezos had from Shaw. Another was a guy named Peter Laventhol. Since Bezos was looking for engineers, Laventhol told him about a friend from grad school at Stanford, a guy named Herb, who was living in California and kicking around his own ideas for an Internet-based business with a friend. And that friend was Sheldon J. "Shel" Kaphan, a well-known engineer who had been bouncing from start-up to start-up in Silicon Valley in search of one that would become the next Apple Computer. It was early 1994.

Kaphan had a B.A. in mathematics from the University of California in Santa Cruz, the laid-back university and surfer town where he grew up. But by the mid-1990s, Santa Cruz, seventy miles down the coast from San Francisco and thirty-five miles southwest of San Jose, had become something of a Silicon Valley suburb. Several technology companies had moved or grown up there, and people (like Kaphan) who preferred the coastal, hippie-ish culture of Santa Cruz commuted to Silicon Valley. Kaphan's reputation as a superb programmer had spread throughout Silicon Valley over the twenty years he had worked there, but he had not yet ended up at a successful start-up where he could make his fortune.

It wasn't from a lack of trying. In 1985 he became senior scientist at Lucid, Inc., an artificial intelligence software company founded a year earlier by scientists from Lawrence Livermore National Labs, Carnegie-Mellon University, and MIT. Kaphan left in 1989, and Lucid went bankrupt in 1994. Kaphan then joined another high-profile company, one-year-old Frox, Inc., as senior software developer. Frox was developing "revolutionary" home entertainment systems, including the much-publicized Frox wand, a one-b.u.t.ton universal remote. Frox had attracted engineers from Lucasfilm, Droidworks, Xerox PARC, Sun Microsystems, and Apple Computer. But Frox didn't have the design flair of Steve Jobs. Kaphan lasted about three years, the company lasted twenty. In 1992 Kaphan joined Kaleida Labs, a joint venture between Apple and IBM, which created a digital multimedia player for computers, and wanted to create a similar program for television set-top boxes. Founded in 1991, it was folded into Apple in 1995.

In the spring of 1994, just about the time Bezos was looking for an idea that he could turn into a good Web-based business, Herb and Shel were doing the same thing. But the pair felt that they knew what their previous start-ups had lacked. "We were primarily technical people and we knew enough to know we needed to link up with a good businessperson," recalls Kaphan. "We knew that whatever we did we would need help with fund-raising, marketing, and management, at least."

Herb and Shel started tapping their own networks to find that businessperson. They reached out to Laventhol, who pa.s.sed the information on to Bezos. Jeff flew out to Santa Cruz to meet them.

The trio met for breakfast at a cafe in Santa Cruz to exchange ideas. They got along well. "I thought Jeff was energetic and primed to succeed, with a history of succeeding," Kaphan wrote to me in an email. "He seemed, when I met him, to be cheerful of disposition and with a good sense of humor. He was also obviously well-versed in the ways of the financial community and I thought that was key to success, as I had worked for a number of places where the management lacked a sufficient degree of skill in that area. It seemed at the beginning that he was willing to trust me to do my job and would give me the s.p.a.ce to do it in the best way that I saw fit, which was important to me." Mainly, he says, Bezos struck him as "someone who was born and bred to be successful."

Kaphan also liked the idea of creating an online bookstore. While less ambitious than the highly advanced technology his previous start-ups were trying to commercialize, it seemed to have a better chance of succeeding. "Having worked for a large number of more technologically oriented start-ups that didn't do so well, I liked the idea of one in which I could easily describe where the revenue stream was going to come from," he recalls. "At that time both Jeff and I believed Amazon could succeed as a relatively small business, compared to what it eventually became. I liked that too."

Plus, it reminded Kaphan of an enjoyable, although brief, time he had spent in 1970 working for Stewart Brand's Whole Earth Truck Store, precursor of the Whole Earth Catalog. "I saw Amazon's mission as a continuation of certain aspects of that same mission: to supply hard to find tools (mainly information-based tools) to a far-flung clientele who might not have easy access to those tools in their local communities," he says.

Bezos offered to hire both Kaphan and Herb. Kaphan even started looking around for office s.p.a.ce in Santa Cruz, hoping that Bezos might decide to put down his entrepreneurial roots there. The conversation between Bezos, Kaphan, and his friend lasted for several months as Bezos made his plans.

Bezos had not yet decided where to base his company. While most entrepreneurs were moving to Silicon Valley as the obvious place to become an entrepreneur, Jeff took a different approach: Yes, he would create another deal flow list to help decide.

He came up with three criteria: It had to be a place with an established population of entrepreneurs and software programmers. He wanted to locate the business in a state with a relatively low population because only residents of that state would have to pay sales tax on the products he sold. He wanted a city near a warehouse run by one of the major book distributors so he could get supplies quickly. But the city also had to be a major metropolitan hub with an airport offering a lot of daily flights so he could deliver books to his customers quickly.

The city that best suited the criteria was not in Silicon Valley. Bezos found that Seattle, Washington-the birthplace of Microsoft-best fit the bill. Because of Microsoft, Seattle was becoming a major technology center, attracting other research centers from tech companies such as Nintendo and Adobe Systems, and was spinning off entrepreneurial companies, including the streaming media company RealNetworks (then called Progressive Networks), started by a former Microsoft executive. In many ways, the greater Seattle area was a younger version of Silicon Valley-and certainly a much cheaper place to live. And Ingram Book Group's largest U.S. distribution center was a short six-hour drive away in Roseburg, Oregon.

Seattle was also the home of Nicholas J. Hanauer, a friend of one of Bezos's former Shaw colleagues. Hanauer was a senior executive at his family's business, Pacific Coast Feather Company. The company provided feathers for pillows, comforters, and mattress pads. In fact, it provided a lot of feathers-enough to bring in $200 million a year in revenues. Hanauer was doing well.

One day, Hanauer decided to visit a friend in Manhattan. The friend invited Bezos along for lunch with him and Hanauer. Although they probably didn't have many mutual business interests, between bird feathers and computer networks, Jeff and Hanauer hit it off.

A couple of years later, when Hanauer heard that Bezos was going to start an e-commerce company, he called him in Manhattan. Hanauer told Bezos that he was interested in investing, that Seattle was "the center of the universe," and that Bezos should move out there to create his new company. That gave an even stronger incentive for moving to Seattle.

But from this point on, the story becomes less certain. The well-known scenario is that Bezos didn't know where they would end up even as he and MacKenzie packed up and prepared to leave Manhattan in the summer of 1994. He was apparently also contemplating sites in Nevada. Another legend about Jeff's fondness for taking bold risks recounts that, when the moving company packing up his Manhattan apartment asked for the destination, Jeff told them to just start driving west, while he and MacKenzie got in a 1988 Chevy Blazer donated by his father and did the same. (It turns out that story is a bit of an abbreviation. Jeff and MacKenzie flew to Texas to get the car, and drove from there.) The day after they left, Jeff made up his mind and called Hanauer, said he was on his way, and asked if he could store his things at Hanauer's house. Then he called the movers with their destination. Jeff and MacKenzie arrived about a week after their furniture.

Kaphan was disappointed that Bezos didn't choose California. "The change of venue from Santa Cruz made it a lot harder for me to decide to do it," he says. In fact, his friend Herb decided not to make the move. But Kaphan decided to throw his programming talents into the pool. "I was convinced to join because I believed in the mission of making a wide selection of books available to people all over the world, because I believed that I could have a significant part in making it happen, and that Jeff was someone who was likely to succeed. I also thought it was going to be fun, and I was tired of the kind of jobs that were available to me in Silicon Valley."

So Kaphan rented a U-Haul to move some of his possessions to Washington. Still, he kept his house and much of his furniture in Santa Cruz, just in case the whole thing failed. Kaphan was Amazon.com's first employee and a key partner in launching the business. Within a couple of years, he made a fortune when Amazon went public. He stayed for five years.

Bezos also had to decide what to call the company. He would need a name if he was going to incorporate a business. On his drive west, he called a Seattle attorney whom Hanauer had recommended, Todd Tarbert, to incorporate the company, and told him the name was to be "Cadabra, Inc.," a play on the magician's chant of "abracadabra." It was apparent that name was a problem when Tarbert asked, "Cadaver?" "I thought, 'Oh, boy, that's not going to work,' " Bezos recalled. When Kaphan moved from California and found out the name of the company he was joining, he said the discovery "almost caused me to return to Santa Cruz."

But Cadabra was incorporated in July 1994, just as Jeff and MacKenzie arrived in Seattle. Seven months later Jeff chose Amazon for a new name because it began with an A, would come up high in any alphabetical listings, and because, as the world's largest river, it reflected his ambitions for his company. Plus, it was easy to spell. "One of the things that people don't think about but is really important is that online, you get to places by being able to spell their name," he noted. But he insisted it always be referred to as "Amazon.com" to show that it was a new breed of business. Amazon became the first successful "dot-com" company, before the dot-com crash turned it into a derogatory term.

Jeff and MacKenzie spent five days at Hanauer's house before finding their own three-bedroom rental, at 10704 NE 28th Street in the Seattle suburb of Bellevue. It cost them $890 a month. He chose it in part because it had one crucial requirement-a garage, so that Jeff could boast of having a garage start-up like Silicon Valley legends from Hewlett-Packard on. The garage had actually been converted into a recreation room, but Jeff figured it was close enough. "We thought it was important for that garage start-up legitimacy," he said. "Unfortunately, it was an enclosed garage, so we didn't get the full legitimacy. But it did have no insulation and a huge, black, pot-bellied stove for heat right in the center of the garage." (The house was later sold, in 1998, for $182,000. In 2009, because of the growth of the technology business in Seattle and the appreciation in the cost of living in the area, it was sold for $625,000. Neither buyer knew its history as Amazon.com's birthplace, although both buyers worked at technology companies.) Ready to start work on his start-up, Bezos needed a business plan, of course. Or did he? Another Bezos legend claims that Jeff wrote his business plan on a laptop in the pa.s.senger seat while MacKenzie drove from Manhattan to Seattle. Others at Amazon, however, have said that he still hadn't finished writing his plan a year after the company was started.

He managed to start a company without a business plan because he didn't need to raise money yet. Ever wonder how entrepreneurs get so rich? The wealthiest-like Microsoft's Bill Gates-manage to build up some value in their company before asking venture capitalists for cash. Gates did that by creating a company that was almost immediately profitable. Amazon.com represented a new approach to entrepreneurship, one that would be used by thousands of dot-com companies in the late 1990s: He went for years without a profit, or even trying to become profitable.

Instead, Bezos used his own money. Having been well paid in his previous jobs, he had enough savings to fund the company for several months. In 1994, Bezos bought 10.2 million shares of Amazon.com stock for $10,000-or one-tenth of a penny per share. When Amazon.com went public three years later at $18 per share, this initial investment from Bezos was worth almost $184 million.

Ten grand, however, doesn't get you far. By the end of 1994 he had also given Amazon.com interest-free loans totaling $44,000. Fortunately, he had friends and family to help out as well. In February 1995 he raised just over $100,000 by selling 582,528 shares to his father, Miguel Bezos, for 17.17 cents per share, enough to keep the company running for the first six or seven months.

His family was willing to invest because they had as much faith in him as he did himself. His mother has said the family invested in Jeff, not in the concept he was pitching, since they had never heard of the Internet before. Bezos knew that only about 10 percent of start-ups succeed even modestly, but gave himself about a 30 percent chance, and told his parents up front that those were the odds. He told his family they should only make the investment if they were prepared to lose it. That kind of honesty is crucial for an entrepreneur, who might otherwise be too intimidated by the odds of failure to make the risky moves necessary to win. "That's actually a very liberating expectation, expecting to fail," he has said. Of course, his family made out very well in the end. That hundred grand worth of stock would be worth over $10 million in a couple of years.

When Jeff got to Seattle, he looked around for other smart software jockeys recommended to him by friends and colleagues. He tapped the University of Washington Computer Science & Engineering Department, which had a great reputation in computing.

Brian Bershad, who headed UW's Science & Engineering Department, was skeptical of the venture when Bezos told him about it. But he circulated an email to people affiliated with his department, and one person decided to take a chance: Paul Barton-Davis. (He reverted to his original name of Paul Davis after his divorce from his wife in 2001, but is now so famous on the Internet by the previous name that it still sticks to him.) Davis was a molecular biologist from London who immigrated to Seattle in 1989 and became a software engineer. At a Seattle company called ScenicSoft, Inc., he designed programs for the open-source UNIX software system, the operating system Amazon.com would also use. In the fall of 1993 he joined the UW Computer Science & Engineering Department, and he became the Web master for the first open Web network (one not restricted to an individual or small group of companies) in the Pacific Northwest.

Davis had aspirations to work with a start-up, however, and didn't last long at the university. The following summer he became a programmer for the Seattle Internet company called USPAN. He created a hyperlink program (which allows Internet surfers to jump from one Internet site to another by clicking on the link) but abandoned it when he discovered the Mosaic Web browser. Mosaic was developed at the National Center for Supercomputing Applications (NCSA) at the Urbana-Champaign campus of the University of Illinois, becoming the first (and, many people still feel, the best) browser to take advantage of the newly created World Wide Web. Davis hoped to integrate his hyperlink program into that browser, and contacted Marc Andreessen, a smart computer science student at the University of Illinois who codeveloped Mosaic. If that collaboration had worked, Davis might have joined Netscape Communications, which Andreessen later founded, and made his fortune there. But Mosaic and Davis's hyperlink system were too incompatible, and Davis had to look elsewhere to join the new wave of bright engineers and entrepreneurs hoping to build new products and companies on the Internet.

So Davis decided to check out this new company starting up in Seattle. He was initially reluctant to join such a raw company, but he was impressed with Jeff and his vision of creating a virtual bookstore online. "I thought he would be interesting to work for," said Davis. He met with Kaphan in October and the two seemed compatible, so he joined the team as employee number two (the first of many former UW engineers Bezos would hire).

Neither Kaphan nor Davis had much experience creating the kind of retailing or business software that Amazon.com needed, but Jeff's philosophy was to hire people with the most talent rather than the most experience. After all, they were trying to do something new, and experience with legacy software could be more of a hindrance than a help. It's a philosophy promoted by Silicon Valley start-ups that the best people are those who don't know that something "can't be done," and therefore will figure out how to do it. Bezos is a strong believer in this philosophy. He and these two men developed the core software that launched Amazon.com, programs that ran the company for years.

The third employee of the company was Jeff's wife, MacKenzie. She handled the phone calls, ordering and purchasing, secretarial duties, and accounting. She had to learn the last job by taking up a simple accounting program, Peachtree PC Accounting Software. Jeff didn't hire a real accountant until the summer of 1996.

At the time, Amazon.com was still a bare-bones garage start-up, with nothing but a desk made from a door and a network router Jeff had bought online from the Internet Shopping Service, no doubt in order to check out an online shopping network. He also checked out the capabilities of the other fledgling online booksellers. He bought How to Be a Computer Consultant from Computer Literacy's clbooks.com.

It was just a few months after arriving in Seattle that Bezos decided to take a cla.s.s on how to create a bookstore. Howorth, the instructor who gave the famous lecture on customer service, remembers Jeff from the cla.s.s. He recalls that Jeff was evasive about his plans. Many of the partic.i.p.ants would have meals together, sometimes talking about subjects they didn't want to raise in cla.s.s. At one meal, Howorth asked Jeff what he wanted to do after the course. "I'm not really sure," he lied. It wasn't that unusual, since many of the students were reluctant to give away information that might tip off a compet.i.tor, such as where they wanted to open their store.

On the last day of cla.s.s, Howorth said goodbye to Jeff. "I said, 'I'm not really sure what you're going to do, but I have a feeling you'll be successful.' " What led him to that conclusion? "I could tell he was bright and quick," says Howorth. "I had no real idea, but the way he was deeply mysterious about it, I could tell he was doing some deep thinking about it, a lot of planning. And he had a very friendly smile."

Howorth had no idea that the smiling Jeff who took his course would become a major compet.i.tor of virtually every bookstore on earth-until a few years later at an American Booksellers a.s.sociation conference. "I saw Jeff and recognized his face, but I was not sure from where," he recalls. "Maybe someone from my hometown? I asked him how he was doing. Then I looked at his shirt, and it said Amazon.com. I suddenly knew exactly who he was. I said, 'Oh, my G.o.d, it's you!' I didn't know Jeff Bezos had been in the cla.s.s."

With an introductory course in bookselling, some experience buying a few items online, one computer, two engineers, his wife, and a garage, Bezos was ready to start building an online bookstore.

Chapter 6.

How to Build a Better Bookstore.

We used to joke that the ideal Amazon site would not show a search box, navigation links, or lists of things you could buy. Instead, it would just display a giant picture of one book, the next book you want to buy.

-Greg Linden, former Amazon programmer.

It took Jeff Bezos and his tiny team just one year to go from settling into Seattle to launching a company. First, he had to get some computers to run his site and store all the data he had collected about books. He bought two or three Sun Microsystem workstations, small but powerful computers often used by engineers to design products or run computer networks.

Amazon's biggest expense was obviously going to be buying or building software. Start-up executives often make the mistake of a.s.suming no outsiders can build software as well as their own programmers. They end up wasting time and money building what they could have easily bought from a vendor that had already worked out the bugs and refined its programs.

Bezos, however, knew what he was doing. He probably knew as much about programming as the people he had hired to do it for him. He carefully chose what to buy and what to build. In order to run the basic operations of the company and manage all the data, he and his programmers chose Oracle Corporation's Oracle database management system. It's an expensive system, but well tested, reliable, and widely used by giant corporations to store and manage their data. In fact, it was more powerful than the tiny company needed at the time. But Bezos had confidence in his future, and Oracle's software would allow the company to grow without having to switch systems. Davis and Kaphan were careful to keep that issue in mind when writing their own software as well. They didn't want to be replacing systems as the company grew, another common problem with start-up companies. "Jeff was very concerned about scaling the company," says Peri Hartman.

Still, Oracle was a general-purpose system. Amazon's software designers had to build upon it to do exactly what the company wanted. In that process, the programmers made a lot of mistakes. Kaphan was primarily responsible for the Web site software that would be seen and used by customers, while Davis focused mostly on the back-end systems that conducted transactions and ran the company. Neither Davis nor Kaphan were experts in relational database systems. "We made some good guesses and a lot of poor ones," Davis would later say.

But the team still had to build a lot of software themselves. Executive hubris? Probably not. This team was building an unusual company of a type that had never really existed before. Since they had to spend time creating their own programs, they did it on the cheap.

In order to keep costs down, they relied heavily on open source software, free programs such as the UNIX operating system and the C and Perl programming languages, all of which were able to run on the Sun workstations Bezos had bought. Open source software is built by hobbyists and university professors and students, is widely distributed, and can be enhanced or added to by anyone. The open source strategy, in fact, became popular for many other dot-com companies, including Google. Although Amazon's programmers had no experience in retail, back-office, or customer-centric software, Jeff got his stock options' worth from them. They were familiar with open source programs, allowing them to create the workhorse software they needed. Davis and Kaphan built their software on top of the free UNIX operating system using the free programming languages.

In addition to the Oracle system that would manage the database program, they had to build their own database that would hold all the information to run the company. So they made it from an open source system called DBM (database manager), which had been created at AT&T and later improved upon by the University of California at Berkeley and others. That required modifying it to make it run on UNIX. When the company launched in 1995, its overall database of books contained more than a million t.i.tles, requiring more than two gigabytes of memory. They put the thousand most popular books into a twenty-five-megabyte computer memory system that could respond very quickly to requests through the DBM program.

With a database containing more than a million books, Jeff began to claim that Amazon was "the biggest bookstore on earth." That was a nice marketing gimmick of questionable veracity. For one, Bezos didn't want his company to have any inventory-or, at least, he wanted the inventory to pa.s.s through his hands very quickly, in and out the same day. What he actually had was a huge database of book t.i.tles and information about them. His plan was to order the books from publishers or distributors only after his customers had ordered them from him (a goal he later had to abandon as Amazon grew). His belief was that Amazon would then be able to operate with much lower overhead than physical bookstores and mail-order companies. Another reason the claim of a million books was an exaggeration is that even the distributors had only about 300,000 books in stock at any one time. In another sleight of hand, although Bezos claimed Amazon to be a bookstore with 1.1 million t.i.tles, his database actually had a list of 1.5 million t.i.tles. One source later told book author Robert Spector that the reason for the lowball figure was so that Amazon could later claim 1.5 million books, making it seem as though the inventory had grown.

In terms of the number of books actually on hand at any one time, Amazon's compet.i.tors could have just as easily described it as one of the smallest bookstores on earth. In fact, any physical bookstore could claim access to just as many t.i.tles as Amazon, because that bookstore could order any books from distributors or publishers just as Amazon did. The difference was that Amazon could find the t.i.tles quickly in its custom-built database and place the orders faster than a physical bookstore could using people staffing a help desk.

Amazon's programmers still had to create a highly customized inventory tracking system, because the company still had to keep track of books moving from the publisher or distributor to Amazon's warehouse and then to the customer. Since mail-order companies did keep books in stock, they typically had only two categories for inventory: in stock or on back order. Bezos wanted more precision, so Davis had to write an inventory tracking system specifically for Amazon. If Amazon already had a copy in its warehouse, the book was listed as able to ship in one day. If a distributor had a book in stock, it was promised in two or three days. (If nearby Ingram had it, the book could often make it out in one day.) If a book had to be ordered from a publisher, delivery to the customer might take a week or two. If the book was out of stock at both the publisher and distributors, it was listed as "shipped in four to six weeks or maybe never." Books that were out of print fell into the "maybe never" category. (Bezos would still try to find the out-of-print books, and sometimes managed to do so from publishers or other bookstores, a feature that gave Amazon a reputation early on as an amazing place to find books.) The general strategy was to be conservative when estimating when a book would be shipped, so that surprises would be positive-shipped sooner than the customer expected-rather than negative.

Still, some of the tasks had to be done by hand and were extraordinarily tedious. That meant a lot of work for Bezos's tiny team of programmers. In order to find books, for example, Amazon used a database of t.i.tles called Books in Print, published by R. R. Bowker, which lists all in-print books by their International Standard Book Number (ISBN). It provided this list-a million and a half t.i.tles altogether-to publishers, bookstores, and libraries on CD-ROM disks. Getting the lists into Amazon's computers was like moving a giant pile of sand with a teaspoon, because Amazon did not have the huge, expensive computers used by large corporations to suck up the data quickly. The programmers could only transfer six hundred books at a time, which meant someone had to copy and paste blocks of t.i.tles twenty-five hundred times to copy an entire disk. R. R. Bowker sent out an updated disk every week, and just transferring the updates took almost an entire day.

Different databases of books were not always reliable, however. They often had different data about which books were available. So the Amazon team determined which were more reliable with a simple test: They would order books and see which databases were accurate. They found that when distributors or publishers claimed that books were out of stock, they were often actually out of print. But when they acknowledged that a book was out of print, that was usually the truth.

The next problem was how to take orders from customers and collect the money from them. That was a new issue for the Internet, which had just recently opened up for commerce. The Amazon brains didn't know if people would be comfortable placing orders and giving credit card numbers over the Internet, or if they would prefer to use email, or to call the company directly, or to simply mail in a check. There were already well-publicized cases of hackers breaking into the files of Internet companies and stealing credit card information. The most obvious solution was to provide for every scenario.

At first, the team thought email was the most likely scenario to work, since email users outnumbered Internet users by ten to one. So they created an email ordering system: A customer could search for a book on the site, but the rest could be handled by email, on the a.s.sumption that people would be more comfortable with it: placing the order by email, getting an email back letting them know how soon it could be shipped, then sending a credit card number back via email.

By the time Amazon opened for business, the petri dish of the Internet had become so popular that the email system wasn't needed, and Amazon was able to rely mostly on orders coming directly through the Web site. About half the customers phoned in their credit card numbers, and some paid by check, but the team was surprised that a significant number of customers were willing to do the transactions directly through the Web site.

Of course, that meant that keeping the credit card numbers safe was of paramount importance. Davis created that system, something he dubbed the "CC Motel," a play on Black Flag's Roach Motel, where "roaches check in, but they don't check out." Credit card numbers could be entered into the system, but hackers would not be able to get them out. The approach was simple: Never put the credit card information into any computer that was connected to the Internet. The card number was transferred to a floppy disk and walked over to the transaction processing computer. (In those early days of unreliable computer networks, companies called this a "sneaker-net" network, a backup for when they couldn't get the network to function properly.) The CC Motel computer was connected by a telephone-based modem only to the credit card companies and to the book distributors, and was used both for charging the credit cards and for ordering the books from the distributors. The books were ordered first, in order to make sure they were available. Only then were the credit cards charged.

This early system worked, although it was almost laughably simplistic, and was one that would obviously have to be replaced once orders reached substantial levels. Davis later recalled that he would have nightmares about the system. The data was supposed to be backed up onto another computer every night, but the team would sometimes forget to do it. Sometimes they lost or accidentally wrote over files that might contain a couple hundred transactions. That required going back to the CC Motel and printing out a list of the credit card numbers, then calling the credit card company to go through the numbers one at a time to make sure the transactions had been processed. At other times, they lost files of credit card numbers and would have to get the credit card company to fax back a list of the transactions, which contained only the last four digits of the credit cards, and someone at Amazon would have to sit down with the list and match them to their list of transactions. "We didn't take seriously the responsibility of keeping that data in a good state," said Davis.

But there were also features of the shopping site that few other companies thought to build in, features that gave customers confidence in using the system. For example, people did not have to even register with the site in order to start looking for books or putting them into a shopping basket. "We let people get well into the ordering process before we made them create an account, which was a real stumbling block on some other early e-commerce sites," says Kaphan.

In order to a.s.suage customers' fears about leaving their credit card information online, Jeff's mandate was to make the ordering process "gentle," as Kaphan puts it. So they were given the option of leaving just the last few digits of their credit card number, and calling in the full number by phone when they were ready to be charged. And Amazon made sure that the customers knew they would not be charged until the very last step, eliminating the fear that they would buy a book accidentally. "Users were always rea.s.sured at each step that they were not making irreversible steps until they were ready to commit their orders," says Kaphan. "I remember that next to the b.u.t.ton to put something in the shopping basket, I put something like 'you can always take it out later.' "

This was all part of Bezos's mandate to make sure the site was the best around, a philosophy that the programmers took very seriously. The Web was new, confusing, and more than a bit intimidating to most people, and generating trust was (and still is) highly valued online. The Amazon.com system helped meet Bezos's goal of creating a good experience for customers. Mainly, he knew that this approach was key to the company's success.

In fact, Davis and Kaphan were not only able to make the site trustworthy and easy to use, they made it more useful than shopping off-line. Any good Web site should exploit the technology of computers and the Internet to do things that can't be done off-line. Very few of the early dot-com companies figured out how to do that. But Bezos reasoned that if customers can get the same service off-line, why change to a new medium that seems confusing or even scary? "The Web is an infant technology," Bezos said at the time. "If you want to be successful in the short-to-medium term, you can only do things that offer incredibly strong value propositions to customers relative to the value of doing things in more traditional ways. This basically means that, right now, you should only do on-line what you cannot do any other way."

Amazon's programmers found a way to offer something that can't be done off-line by making it easy to tap the company's database in order to find information about books and authors. Kaphan tapped his expertise in creating hypertext, or links from one piece of data to another. He and Davis took all the biographical information about authors in the Amazon database and intertwined it through links. Customers were able to search through all subjects and authors they found of interest. If they found one book they liked, they could click on the author's name and find all the other books he or she had written. They could click on a subject category to find other books on that subject. Who needs a Dewey Decimal System? "I always thought that the hypertextualization of the bibliographic information was key [to Amazon's early success]," says Kaphan. "You could navigate through the large s.p.a.ce of available books."

Aside from weeding out the bugs in the internal workings of the site, Amazon's programmers had to make sure the site would work properly with several different Web browsers, each with different features. The graphical browser was what made the Internet popular, by turning it into a simple point-and-click navigation system. In the early days of the Internet, n.o.body could be sure which Web browser would be most popular.

University students from around the world were cranking out graphical browsers to take advantage of the World Wide Web communications standards developed by Tim Berners-Lee at the European Organization for Nuclear Research in the suburbs of Geneva. Some of the earliest graphical browsers from the early 1990s were now largely forgotten or overlooked names like Erwise, developed at the Helsinki University of Technology; ViolaWWW, from the University of California, Berkeley; and Lynx, created at the University of Kansas.

But the one that really got people moving to the Internet was Mosaic, developed at the Urbana-Champaign campus of the University of Illinois, in the school's famed National Center for Supercomputing Applications (NCSA). It was the foundation for the Netscape Navigator, the first successful commercial browser (although offered for free), created at Netscape Communications by former Illinois students who had worked on the Mosaic browser. By 1995, Netscape also had compet.i.tion from Microsoft's Internet Explorer, which ended up capturing most of the market.

By the spring of 1995, Amazon had a Web site. The site wasn't finished yet, but it was working well enough to let a few hundred friends try it out-after they were sworn to secrecy, of course. With the promise not to tell anyone about the project, the beta testers began browsing through books and making pretend purchases.

One problem Amazon's programmers discovered during testing was that there was no way to track an individual customer's activity. If someone bought a book, then started browsing for other items, Amazon's computers had no way of knowing that both actions were done by the same person. It's one of those things you don't realize you need until you discover you don't have it. They had to build in a system that would store all activity from a particular user into one file, and pull out that data when a customer returned.

Some of the site's important features just reflected a lack of resources. The programmers used few graphics in the early days, mainly because they didn't have many to work with. They were not graphic designers, and many book publishers were either unable or unwilling to supply them with images. But that turned out be a feature customers liked. Internet connections were very slow in those days, and graphics took a long time to download. Also, some of the early Web browsers were text-only, unable to display graphics, so Kaphan and Davis made sure all the necessary information was provided without the need for images. Just try turning off the graphics in a Web browser today and you'll see how many critical pieces of information and links disappear from most sites.

Again, many early dot-com companies missed the point. They put as many graphic images as possible on their sites (and many still do) under the mistaken belief that it makes their sites look more professional or their ads more noticeable. In reality, the sites just become more annoying and confusing. In fact, Google was later to stumble on this same approach, demonstrating that a spare Web site is still vastly superior to one flashing images and shouting out sound effects like a late-night TV pitchman selling $19.95 items not available in stores.

Bezos had to improvise to deal with unexpected problems, and he did so with elan, if not a devious glee. For instance, he wanted to test his ordering systems to make sure they worked properly and to work out bugs. That meant ordering one book at a time. But the distributors wanted a minimum order of ten books. He was later to say in his speeches that he had offered to pay the distributors extra to get just one book, but they refused to break procedure. Bezos discovered, however, that if some books in a particular order weren't in stock, the distributors would still ship the others and charge for only the books shipped. He then found an obscure book on lichens that both distributors supposedly carried, but did not actually have in stock. So in order to run his tests, he started placing orders for the one book he wanted, plus nine of the lichen books. "They would deliver the one that we wanted, along with a very sincere apology about not having been able to fulfill the nine copies of the lichen book order," he said. "That worked very well for exercising our systems. I've since talked and joked at length with the people at these companies about this. They actually think it's very funny."

In many ways, Amazon's early design was a combination of trial and error, good guessing, chance, and some clever improvisation. But the small team kept an eye on what was important: giving priority to the customers' needs rather than trying to extract every dollar possible from their virtual wallets. They paid attention to what customers liked and what they didn't. That approach gave Amazon a good start-and continued to be a guiding beacon in the company's future.

Chapter 7.

Growing Pains.

Amazon.com was finally launched on July 16, 1995. It was just in time-just as ma.s.ses of people started moving onto the Internet and before many compet.i.tors had created good commercial sites. Amazon.com was launched with the latest technology and cleanest design in the middle of the year when Internet use grew several times, to sixteen million people. Bezos was able to take advantage of his observation of bacteria like growth online to get an early start on compet.i.tors. That fortuitous timing was a key piece of luck and hard work from the programmers.

Bezos was also able to move his company out of his altered garage. He set up shop in an industrial neighborhood in Seattle that he shared with a needle exchange and a shuttered p.a.w.nshop. He had eleven hundred square feet of office s.p.a.ce on the second floor and four hundred square feet in the bas.e.m.e.nt to be used as a warehouse. The desks were made from cheap doors with sawed-off two-by-fours for legs. (As Amazon grew, Bezos hired a carpentry firm in Seattle to build them for $130-$70 for materials and $60 for labor.) One room of the office s.p.a.ce was used to store cardboard boxes. The CC Motel and Internet computers sat on a cheap metal shelf. The warehouse held more shelves to store just a few hundred books on their way from the distributor to the customer, a couple of tables with supplies to pack up books for shipments, a metered scale, and a Pitney Bowes postage machine. He also placed whiteboards in the elevators so employees could scribble ideas during some of the only free time they managed to squeeze into their days.

Orders started coming in as soon as the site launched. In the first few days, there were half a dozen orders per day. One of the programmers set up a program so that every computer at Amazon would ring a bell every time an order came in. A great novelty at first, it quickly got too annoying and had to be turned off after a few days.

Discounted books, of course, were the first big draw to Amazon's site. When the site first launched in 1995, Bezos put everything on sale. The top twenty or so best-selling books were sold at 30 percent below list price, which meant Bezos was selling them as loss leaders. This is a practice often used by the big bookstore chains. (Perhaps Bezos learned that in his bookselling cla.s.s.) But Bezos also cut the price of all three hundred thousand t.i.tles that the distributors had in stock by 10 percent.

The staff picked out a special selection of books for the daily Spotlight, chosen partly by how much information was available to add as a Spotlight profile, and discounted them by up to 40 percent. With these discounts, Amazon could beat the prices of even the big chain stores.

The fact that Bezos managed to get online before the big bookstore chains was an invaluable advantage. A huge number of those early Internet users were the famed "early adopters" of new technology that all tech companies covet. The early adopters found Amazon.com, decided it was good, and word spread throughout the Internet. After that, it was hard for any other bookseller, no matter how good, to match Amazon's cachet.

Three days after launch, Bezos got an email from Jerry Yang, one of the founders of Yahoo. "Jerry said, 'We think your site is pretty cool; would you like us to put it on the What's Cool page?' " Bezos recalled. "We thought about it some, and we realized it might be like taking a sip from a fire hose, but we decided to go ahead and go for it." Yahoo put the site on the list, and orders soared. By the end of the week, Amazon took in over $12,000 worth of orders. It was hard to keep up. That week, the company shipped just $846 worth of books. The following week brought in nearly $15,000 worth of orders, and the team was able to ship just over $7,000 worth of the orders.

The site wasn't even truly finished when the company launched. Bezos's philosophy was to get to market quickly to get a lead on the compet.i.tion, and fix problems and improve the site as people started using it. One mistake they discovered after launch: People could put a negative number into the box that asked how many copies of the book they were ordering. "We found that customers could order a negative quant.i.ty of books!" Bezos later recalled. "And we would credit their credit card with the price and, I a.s.sume, wait around for them to ship the books."

Another mistake was that Bezos had not hired anyone to pack up and ship the products as they arrived from the distributors. The first few weeks, everyone at the company was working until two or three in the morning to get them packed, addressed, and shipped.

Bezos had also neglected to order packing tables, so people ended up on their knees on the concrete floor to package the books. He later recalled in a speech to Lake Forest College that, after hours on his knees packing up books, he commented to one of the employees that they had to get knee pads. The employee, Nicholas Lovejoy, "looked at me like I was a Martian," Bezos recalled with a laugh. Lovejoy suggested the obvious: Buy some tables. "I thought that was the most brilliant idea I had ever heard in my life!" Bezos finally went to the Home Depot and splurged on some tables.

The company was strictly a cut-rate operation. Any printing or copying was done at a Print Mart shop a few blocks away. Business meetings were held at a local coffee shop-ironically, one inside a Barnes & n.o.ble bookstore. Employees bought the office supplies they needed and submitted an invoice at the end of the month to be reimbursed. All employees, including Bezos, answered emails from customers with questions, comments, or complaints.

Despite what seemed to be a pathetically amateurish operation, Amazon grew up very quickly once it was launched. By October, Amazon had its first day logging in one hundred book sales. In less than a year, it had its first hour with an order of one hundred books. Word kept spreading, despite the fact that the company did no advertising its first year-although Bezos did hire mobile billboards to cruise by Barnes & n.o.ble stores displaying the question, "Can't find that book you wanted?" along with Amazon's Web site address.

Word spread through the Internet. Netscape put the site on its "What's New" list, and many other sites did the same. Amazon was always near the top of the lists because the lists were mostly alphabetical-exactly why Bezos chose a name that began with an A. Even The Wall Street Journal ran an article about the tiny company on its front page. That got traffic soaring. "It was also a curse, because it alerted our compet.i.tors-of which there were many and large-to our existence," Bezos recalled. "In particular, Barnes & n.o.ble started to take notice of what we were doing." About a year later, in May 1996, Barnes & n.o.ble launched its online bookstore, about a week before Amazon's initial public offering.