← all stories

After the Web · Era 3 · The early Web

1994–1997IPO on Nasdaq · 1997

Amazon

An online bookstore run out of a house in Bellevue that nearly kept the name Cadabra. Orders went to a wholesaler, and the customer's money arrived before the supplier's invoice: at the end of 1996, inventory of $0.57 million against accounts payable of $2.85 million. Revenue grew from $0.5 million (1995) to $147.8 million (1997); in May 1997 the company went public on Nasdaq while still losing money, and later, in 1999–2000, it opened its site to third-party sellers. A by-product was the self-service Associates affiliate program (1996).

Founders Jeff Bezos (the only officially recognized founder) · Shel Kaphan (the first employee: he agreed with Bezos on the job before the company was registered and arrived in October 1994; he describes himself as a co-builder, formally not a co-founder)
Domains amazon.com · relentless.com (registered earlier as one of the candidate names, September 1994; still redirects to amazon.com)
ecommerceaffiliatereferralmarketplaceretaildistribution-partnershipslogistics

In brief

Amazon began as a bookstore with almost no inventory. Most orders were filled by shipments from a wholesaler, and the customer's money arrived before the supplier's invoice. The company grew from $0.5 million in revenue in 1995 to $147.8 million in 1997 fact and went public on Nasdaq in May 1997 while losing money, losses its S-1 said would continue. A by-product was the Associates affiliate program. Less than two years after the IPO, the same Amazon opened a marketplace for third-party sellers on its own site, that is, it moved into the territory where eBay had been operating since 1995.

How it started (the founders)

Jeff Bezos worked at D.E. Shaw, a New York fund. He explained his 1994 decision to leave with a "regret minimization framework": at 80 he would not regret having tried. From a list of product categories he chose books, for their low price, the huge catalog, and the ready-made database of the wholesaler Ingram estimate. Washington state offered one more advantage: sales tax was charged only to residents of the state, and Ingram's warehouse was nearby in Oregon estimate.

The company was incorporated on July 5, 1994, as Cadabra, Inc. The first employee was the programmer Shel Kaphan, who arrived in October 1994, and Paul Davis joined a little later; the office was in a house in Bellevue, and Bezos's wife MacKenzie kept the books. In November 1994 the lawyer Todd Tarbert heard "cadaver" in the name, and the company was renamed Amazon.com. Bezos had registered the relentless.com domain earlier, in September 1994, and he still redirects it to amazon.com. The exact registration date of amazon.com (November 1, 1994) is confirmed only by a school newspaper, so it remains an estimate.

The starting money came from family and acquaintances, not from venture funds. According to press accounts, Bezos held about 60 meetings to raise ~$50,000 a person; 20 agreed (CNBC) or 22 (Yahoo Finance, SCMP) estimate. His parents put in $245,573 estimate. The record gives a third number: per the S-1 section on private placements, 23 investors bought 3,021,000 shares for $1,007,000 in a series of purchases from December 6, 1995 to May 16, 1996; another $345,525 came from three investors in 1995–1996. On June 21, 1996, two investors, among them entities of Kleiner Perkins, put in $8,000,014 for preferred stock. By the end of 1996 the company had raised $1.6 million in common stock and $8.0 million in preferred stock.

Year-by-year timeline

Legend vs. the record

  1. "Bezos started in a garage with no money." The garage is the company's own version: a 2000 press release dates the launch to his garage, while the first employee, Kaphan, describes an office in a house in Bellevue. But "with no money" is wrong: per the S-1, by the end of 1996 the company had raised $1.6 million in common stock from private investors, plus $8 million in preferred stock from two investors, among them Kleiner Perkins. Verdict: the garage is in the company's own account, "with no money" is wrong; venture money arrived only in June 1996.
  2. "Amazon went years without a profit because it could not make one." The S-1 shows a choice, not helplessness: in 1996, $6.09 million went to marketing against gross profit of $3.46 million, and the net loss was $5.78 million. The 1997 letter to shareholders explicitly puts cash flow ahead of attractive GAAP reporting. A caveat: a 22% margin really did leave little room, and even without advertising the profit would have been small estimate. Verdict: a decision rather than a necessity, but on a thin margin.
  3. "Amazon was the first online bookstore and the first affiliate program." Book Stacks Unlimited had been selling books online since 1992, three years before Amazon estimate. CDNow launched an affiliate program with commissions in November 1994, according to industry reviews estimate. Amazon itself wrote in 1998 that it had become the pioneer in July 1996; that is its own claim. Verdict: not the first; the first to make it self-service and mass-market estimate.
  4. "The flywheel diagram and the phrase Get Big Fast were born in the 1997 letter." Neither appears in the full text of the letter, which was checked line by line. The flywheel is linked to a meeting with Jim Collins around 2000–2001 estimate. Verdict: a myth.

Lesser-known but significant facts

  1. Accounts payable five times the inventory. As of December 31, 1996, inventory stood at $571,000 and accounts payable at $2,852,000. Ingram accounted for 59% of purchases, and in the S-1 the company writes that it keeps minimal inventory and relies on rapid delivery from wholesalers, with no guarantee of payment terms. The S-1 does not use any term for a negative cycle; it is an inference from the balance sheet on a single date estimate.
  2. A lawsuit three days before the IPO. Barnes & Noble claimed that Amazon was a book broker: its Seattle warehouse held only a few hundred titles, and the rest came from wholesalers. The suit was settled with no payment and no admission of fault.
  3. Desks made from doors. In the summer of 1995 Bezos bought a door at Home Depot instead of a desk because it was cheaper. That is how employee #5, Nico Lovejoy, tells it, and a second employee, who joined a year later, recalls how wobbly those desks were. Bezos himself also told the story of the kneepads on the concrete warehouse floor: an engineer suggested buying packing tables, and productivity, according to Bezos, doubled. The reading of this as a symbol of frugality belongs to the company itself; it is the company's interpretation estimate.
  4. The "first book," with a caveat. Wainwright's purchase on April 3, 1995, means little in itself: Smithsonian writes that it has not been fully confirmed that he was the buyer estimate.

The first growth lever

Growth came through other people's audiences, and the suppliers provided the working capital. First came free attention. In 1995 the site made the Yahoo and Netscape lists, and in the first month orders came in from 50 states estimate. A one-time stroke of luck could not be repeated on demand, so in July 1996 Associates launched: a site owner puts up a link to a book and earns a commission on the sale, while Amazon handles the order, the payment, and the delivery. The first growth figure was more than 4,800 participants as of December 31, 1996. After that, per the company's press releases, came more than 15,000 sites in September 1997, 30,000 in February 1998 with commissions of up to 15%, and 60,000 in June 1998. Neither the S-1 nor the releases name the starting rate; 3–8% appears in only one secondary source estimate.

What happened next is less well documented. The list of partners in the 1997 letter (America Online, Yahoo!, Excite, Netscape, GeoCities, AltaVista, @Home, Prodigy) is confirmed. But the claim that Associates took off precisely after the paid partnerships with Yahoo and AOL comes from only one reference source, and the letter itself does not connect the partnerships with Associates estimate. So the thesis of a free mention first and a paid partnership after it remains a hypothesis.

The paired story

The pair is eBay (AuctionWeb). Both companies grew out of the same year, 1995, but on opposite models. Amazon bought books from Ingram, kept the order and the delivery in its own hands, and earned the price difference, a gross margin of about 20–22%. eBay began in September 1995 as an auction section on Pierre Omidyar's personal site, ran free for about five months to reach critical mass, and introduced fees in February 1996; it had no warehouse and no goods. Revenue for 1996: Amazon $15.7 million with a loss of $5.8 million against eBay's $372,000 with a profit of $148,000. For 1997: $147.8 million against eBay's $5.7 million with a profit of $874,000. Both companies went public on Nasdaq at $18 a share, Amazon on May 15, 1997, and eBay on September 24, 1998.

The main contrast is in working capital: Amazon built its growth on accounts payable to wholesalers and paid for it with losses, while eBay put no money into inventory at all. The tail of the story is that Amazon itself moved onto marketplace territory with auctions (March 1999), zShops (September 1999), and Marketplace (November 2000). Marketplace put third-party sellers' offers on the page of the same new product, that is, it used the audience the store had already built up. Why Amazon did not stop at the warehouse alone, the documents do not say unverified.

Parallels today (projects from the catalog)

What a builder can take from this in 2026

  1. Other people's working capital can serve as fuel. Amazon held $0.57 million of inventory against $2.85 million of accounts payable; the risk is that nothing guaranteed the wholesalers' payment terms.
  2. A thin margin makes losses taken by choice fragile. With a 22% gross margin and $6 million spent on marketing, the company needed the capital markets; before the IPO Amazon managed to raise $8 million from two investors.
  3. One-time attention has to be turned into a channel. The Yahoo lists led to Associates; the first push could not be made repeatable at will estimate.
  4. Models do not exclude each other. A warehouse and a marketplace share one audience; in 1999–2000 Amazon added a marketplace to the store rather than replacing one with the other.
  5. Do not take "first" on faith. It was neither the first online bookstore nor the first affiliate program; what set Amazon apart was scale and self-service estimate.

Discrepancies and what we could not verify

Sources (primary first)

Primary:

Secondary:

Internet History Keeper

Every historical dossier here is free and open, and it will stay that way. If you want the series to keep going (new dossiers, checks against primary sources, the English edition), support it with a subscription for $9.90 a month. It is support, not access, and you can cancel at any time.

Become a keeper — $9.90 a month →

The same thing, about today

The same breakdowns, but of projects launching right now: what the product is, where the first users came from, how they charge. The card is free, the full dossier is $5 (the dossier itself is written in Russian).

Browse the dossier catalog →