On the Internet, bigger is much better. In a study of the market valuations of US business-to-consumer (B2C) Internet companies, McKinsey found confirmation of ever-increasing returns to scale in their share prices. Visitors to the 19 content sites (in the McKinsey sample¹) with fewer than five million visitors have an average market value of $130, for instance. But visitors to sites with from five million to ten million visitors have an average value of $170, and visitors to sites with more than ten million visitors are worth $460 (Exhibit 1).² Whatever the absolute value of the Internet companies studied, their relative valuations have maintained the same relationships to one another and to the underlying metric—the number of visitors at each site—even amid NASDAQ’s recent gyrations.

Our regression analysis of 21 electronic retailers shows that they too manifest this phenomenon. Analysis also shows that the valuations of e-tailers that have smaller market shares than their competitors do are lower than their revenues and the number of their visitors would otherwise suggest. The market’s expectation of increasing returns to scale is reflected as well in the concentration of capital on the World Wide Web: just 5 of the 192 companies in the Bloomberg US Internet index account for one-third of its $1 trillion value.

The impact on value of increasing returns to scale dwarfs that of more conventional indicators. Many analysts have speculated that profit and market liquidity, or the lack thereof, play a deciding role in the valuations of Internet companies. Yet neither metric can explain the variations in their market capitalization after we had accounted for the influence of numbers of visitors and revenue (Exhibit 2).

In well-behaved traditional industries, returns to scale eventually tail off. Why should the Internet be any different? Is it just the latest version of the tulip craze in 17th-century Holland? The answer might emerge from efforts to think about the concrete advantages that scale brings to the Internet.

Some Internet sites are exchanges that aggregate buyers and sellers of a particular class of commodity: eBay, for instance, hosts auctions in which consumers bid for goods that businesses or other consumers want to sell; Priceline.com hosts reverse auctions, in which businesses decide whether to accept bids from consumers. Regardless of the type of site involved, more buyers attract more sellers, who in turn attract more buyers. Thus, eBay’s market share continues to climb despite the entry of Yahoo! and Amazon.com into the on-line auction market—and despite eBay’s higher fees and commissions (Exhibit 3).

The leaders of nonexchange retail segments are also valued more highly than are their smaller rivals: eToys, for example, is worth $500 per visitor, SmarterKids.com only $30. Amazon and Buy.com are both broad-category retailers, but Amazon customers have twice as much value as do Buy.com customers even though Amazon customers spend much less. At first glance, the network effects at auction sites don’t appear to be at work at other kinds of retail sites. How then can the increasing returns to scale be explained?

One possibility is that leading sites lock in first-time users more efficiently than their competitors do. The more popular sites generate greater brand name recognition, so they are the first sites many people try. If such sites provide a satisfactory experience, the user may be reluctant to look elsewhere. Another possibility is that after taking the time to fill in the information requested by a site, users hesitate to do so again at others.

In addition, the larger a site’s database on its current customers’ choices, the more accurately that site can predict the preferences of new customers and even (like eToys) guide their purchases. Indeed, there may be a virtuous circle in which more visitors to content sites generate more advertising revenue, some of which can be spent on generating better content to attract still more visitors. Size also makes a content site more attractive as a partner in an alliance in which each party directs traffic to the other.

Finally, the tendency of an exchange site’s value for individual users to increase with the overall number of users on that site (such as eBay and Priceline) may also apply to nonexchange sites, albeit somewhat more subtly. Yahoo!, for example, has a messenger function (similar to America Online’s ICQ function) that alerts users if their acquaintances are on-line, so that live interactions can take place. Such functions grow in value as the chance of these encounters increases along with the number of users.

Collectively, such effects create advantages for larger players. What might these advantages mean for managers thinking about Internet business? Increasing returns to scale ultimately allow a single large player to overwhelm all competitors. If this tendency applies across the Internet, it is worth doing almost anything to gain scale; every extra visitor makes all existing visitors more valuable. If the market is right, and returns to scale keep getting bigger, the largest player in a sector is likely to have much higher profits than its competitors do. Assuming that there will be room for pure-play Internet companies in the future, significant spending today may be justified if it marginally increases the chance of winning the race to scale tomorrow.