Even Amazon.com did not fully appreciate all the factors driving its cost-to-serve. When Jeff Bezos opened his Internet store in 1995, he started with books, reasoning that it would be easier to offer the millions of titles in print online than through a traditional mail-order catalog. Although Mr. Bezos may not have fully grasped all of the inefficiencies inherent in the book industry, he saw that his model could minimize inventory risk — a significant problem in book retailing. Unlike other manufacturers, publishers take back all unsold copies of their product. Up to 30 percent of trade books ship back to the publisher at enormous cost to everyone: to the publisher, who refunds the payment to the retailer; to the retailer, who pays for shipping and restocking; and indirectly to the consumer. Amazon’s online model minimizes the inherent inefficiency of placing potentially unsellable titles on thousands of retail bookshelves by relying instead on a relatively small inventory to support a “virtual bookshelf.” If Amazon returns fewer books than a bricks-and-mortar retailer, it should be able to negotiate lower prices from the publishers (reflecting its lower cost-to-serve as a customer of the publisher). Amazon can then pass along those savings through lower prices to the consumer.
Although books worked for Amazon, toys were a different matter. As Jeff Bezos learned, the toy supply chain comes with a much higher cost-to-serve. Toys are more seasonal than books and demand for them is far less predictable. Magnifying those challenges, most toys are made in Asia, and the replenishment cycle can easily outlast the actual selling season. That means merchandisers must accurately predict which toys will be hits and then buy enough inventory for the whole season; Amazon therefore gained no benefit from its “virtual shelves.” Guessing too conservatively results in missed sales, and guessing too optimistically leads to write-offs because toy manufacturers do not typically accept returns of unsold goods. Without any toy merchandising expertise, Amazon guessed wrong for the 1999 holiday season and wrote off $39 million in excess toy inventory in early 2000, having sold only $95 million worth of toys.
Furthermore, conventional toy retailers enjoy advantages that traditional booksellers don’t. Unlike Barnes & Noble, Wal-Mart and Toys “R” Us ship multiple truckloads of goods to each store weekly, which puts their transportation costs far below the cost of shipping individual toys to consumers. Recognizing these differences, Amazon gladly partnered with Toys “R” Us in 2000, shifting the inventory risk to the experts but leveraging the additional product lines to lower its own shipping cost for multi-item orders.
Nonetheless, the two partners had a falling-out in 2004, with dueling lawsuits in the New Jersey court system. This denouement suggests that the cost-to-serve for online toy retailing produced a financial model that would not support the two parties’ aspirations adequately.
Channels and Brands
Although successful online retailing depends on a variety of factors, the cost-to-serve of a particular product category can explain much of the variance in the penetration rates of Internet sales. For example, according to the United States Census Bureau, in 2002 (the latest data available), 44 percent of retail sales of computer hardware and software took place via the Internet. Certainly tech-savvy computer buyers are more likely than the general population to shop online. But equally important, the high value-to-weight ratio of computers — especially as laptops become more popular — minimizes the importance of transportation cost and makes the category a low-cost-to-serve option over the Internet.
Cost-to-serve factors like inventory, packaging, shipping, and returns help explain why online sales of books outstrip online sales of other kinds of merchandise. E-commerce accounts for nearly 13 percent of the total bookstore and newsstand sales in the United States — $15 billion in 2002 (the latest year available from the U.S. Census Bureau) — and the same portion of the $21 billion market for office equipment and supplies. By contrast, the Internet accounted for just 2.7 percent of the $92 billion in U.S. sales of furniture and home furnishings. Or consider the largest retail category, food and beverages. Online sales accounted for only two-tenths of 1 percent of the $450 billion U.S. food and beverage retail sales total in 2002. As Webvan (and many investors) learned, targeting a huge market does not guarantee success if the cost-to-serve economics don’t work.