Conference Summary

The Art and Science of Market Platforms
What Bloomberg, eBay, Nokia and Yahoo! Personals Have in Common
New York City - February 13, 2003

Marco Iansiti- Harvard Business School
What explains the difference between Visa and the Global Trading Web? Why did eBay grow to be a household name, managing hundreds of billions of dollars of transactions, while Enron’s markets went down in flames? Much of our economy is being driven through complex marketplaces that organize the activities of large communities, in industries ranging from software to media, and from banking to energy.
Complex marketplaces are not a product of the “new economy.” While the emergence of distributed information technology has provided opportunities for increasing the impact of these markets and decreasing their operating costs, many of the principles behind the operation of “multi-sided” markets have roots in traditional, established businesses, ranging from dating services to credit cards, and from apparel distribution to securities trading. 

The Global Trading Web, eBay, American Express and Enron all manage multi-sided markets that connect two or more disparate but interdependent groups of customers. American Express, whose credit cards link “communities” of retailers and consumers, is the classic example. The payment card would offer nothing to retailers unless shoppers carried the cards. Similarly, Amex cards would be worthless to shoppers if retailers didn’t accept them for payment.

Not all multi-sided markets fit the dimensions of a traditional market place––software operating systems, like the ones sold by IBM, Microsoft, Palm and Sun, serve market-like functions in the sense that they must attract thousands of application developers to satisfy the needs of millions of computer users. Trading companies like Li and Fung aggregate more than 8,000 companies and 1,000,000 employees to provide the comprehensive supply chain services for large apparel retailers like the Gap. 

In the 1990s both academics and business managers too often lived by the faith that once markets were created, the rest would take care of itself. Yet activities like customer acquisition, transaction coordination and business-process integration are essential to value creation and liquidity. And these tasks require hands-on management through the organization of what we call a “market platform.”

In a multi-sided market, prices must be set so that the relevant communities are attracted to the market in appropriate numbers. In some cases, enough is known about demand and costs characteristics to use mathematical models to determine optimal pricing. More typically, effective pricing requires experimentation––and non-intuitive solutions. For example, with software operating systems, prices are set so that most of the cost burden is carried by end-users––not by developers who create software applications.

Elements of an operating strategy for a market platform include core operations, integration, and coordination. And the failure of many business to business (B2B) exchanges provides a plethora of examples of ineffective operating strategies. Thus while SupplierMarket, a B2B exchange for direct materials, attracted thousands of users to its site and listed a large number of transactions, it never generated adequate liquidity. With weeks needed to close a transaction, the system could not reach adequate scale and the marketplace failed.

Successful market platforms offer a stark contrast. Yahoo Personals boasts continually updated software with a sophisticated matching algorithm to maximize the operating effectiveness and the liquidity of the site. It integrates closely with other Yahoo! properties and is built around a clever promotion strategy that helped the service to reach critical mass in a balanced fashion.  

The Visa payment card cooperative developed clear rules and procedures, including an ownership structure that has accommodated both rapid growth and rapid changes in membership, an agreement to charge uniform fees for clearing transactions between merchants’ bank and card issuing banks, and unambiguous rules for apportioning risks associated with fraud and non-payment. The platform grew rapidly, even as costs declined and service quality improved.

Market platforms are increasingly shaped by decisions about technology. In software operating systems, the core technology enables the fundamental functions performed by a computer, while in bond trading it is aimed at risk management. It is important to note that the difficulty of developing this kind of expertise in many unrelated markets simultaneously––as Enron attempted to do.

Technological integration is increasingly important, too, in everything from software operating systems like Windows to trading systems like Li and Fung. Li and Fung (along with other major supply chain platforms like Wal-Mart and Dell) offers interfaces to connect their systems to individual enterprises’ procurement and management software. In some cases, it will even pay for the integration.

Coordination is the third critical dimension of technology strategy for a market platform. Effective market platforms require heavy investments to ensure that standards and interfaces are well understood by users. When Intel developed MMX technology for its Pentium processors during the mid 1990s, it began informing its community of third-party developers on how to use the technology two years before the first MMX chips were unveiled.

Some lessons seem to hold for platforms across a diverse range of industries:
Liquidity first: One common source of failure is making large investments in technology and operations early, before market liquidity was achieved. The most successful platform owners often are among the most cautious. Both Microsoft and eBay were cash-flow-positive from the start. Both became pervasive in small communities before any significant investment had been made.

Platform flexibility: With early market liquidity the priority, platforms will almost require substantial changes as they grow. To the extent possible, the design should be componentized so that old capabilities may be combined with new ones in an ever-evolving architectural framework. With Microsoft’s platform, what began as a basic software compiler carried around on paper tape has evolved through DOS and Windows, and is now being extended through the .NET framework.

Select new markets through experimentation: Enron encouraged employees to lock the company to target markets. eBay’s platform, by contrast, is designed to follow the user base wherever it goes. No commitment is made until a niche market’s strength and durability has been proven.

Leverage external communities to minimize internal operations: A well-designed platform should minimize the internal effort needed to make certain that transactions clear. In platforms as different as eBay and Windows, the design of a comprehensive, easy to use tool kit that gave participants considerable initiative was essential to success.

Create trust among participants: Early on, eBay’s founders tracked sellers personally, interacted with them in chat rooms, and developed an informal trust rating. A trust management system is still a fundamental part of its platform, although a distributed rating system has naturally taken the place of the founder’s e-mails.

By the same token, a platform should exhibit a clear governance structure. Some successful platforms (e.g., Microsoft, American Express) are publicly traded companies, while others (e.g., the New York Power Pool, Visa) are nonprofit cooperatives. Both extremes work; what seems not to work are hybrids.

Michael J. Reuschel-Unisys Corporation
Everyone has an opinion today about how to revive the very troubled telecommunications troubled industry. I can’t claim to have all the answers, but do have some observations.

If you build it, they won’t come. In two years, telecom stocks have fallen by 40 percent and employment has been halved. Why? The industry made huge investments in broadband, only to be confronted with a glut of providers, a declining pool of demand and a lack of content. Today, less than 10 percent of all of the fiber in Europe is being used, while in the United States we use just one-third of capacity at peak hours.

By the same token, the industry invested $115 billion in “3G” wireless capacity, only to find a lack of customer interest and technology needed to transport 3G dta. Now they’re doing it again, investing heavily in 2.5G services and Internet telephony without evidence the bet is worth making. All told, the industry spent more on building network infrastructure in the last decade than the GDP of India; the only industry with a worse return on equity is airlines.

It’s Time to Milk the Cash Cow. The basic voice business is still hugely profitable for carriers––and supports their mountains of debt, which now equal to 88 percent of industry. But competition is even eroding voice revenues.

The substitution of wireless for fixed wire calls threatens this even more in the long term, and nobody knows what Internet phones will do to call volume. However, some things can be done to milk cash from the voice commodity––adding new features, looking beyond First-World markets, looking ahead to Internet phones.

It’s the customer, stupid. Technology has also always driven customer interest. The latest example is Bluetooth, which was developed in 1994 but has experienced slow adoption. Prices for chipset are high; compatibility is in question; big companies like Nokia and Microsoft have wavered in their support.

Instead of investing in new infrastructure, this is the time to invest in understanding what people want–and what they are willing to pay for. Customers are not being given reasons for spending money on new services. Case in point: nobody’s explaining why customers should pay for expensive broadband.

One result: customer churn has become a huge problem. New promotions seem to be focused on buying market share rather than on delivering unique features or better service.

Horizontal or Vertical? Most telecoms think of themselves in the voice telephony business. But they need to think of themselves as players in the content transportation business, which can include third-party content. That argues for rethinking business models. Should they adopt a more traditional vertical structure––which is the way most telecoms view their business––so that they own it all? Or, should they go the DoKoMo I-mode route, in which brokers’ content and provides the infrastructure to carry and bill for it.

There are lots of questions here. How will telcos manage relationships with third parties? What sort of pricing works? Who will pay for infrastructure? What is the role of regulation?

It will not fly without ROI. The concept of free cash flow rather than return on equity is driving decisions today in the telco world. The more realistic path forward is to evaluate ROI and profitability on the basis of what different aspects of the business are doing.

Where lies the future of telecoms? Less in finding the next killer application and more in using the killer application under our noses. In its broadest form, content is really all about messaging. The trick is to create a platform that allows these messages to be created, managed, stored, retrieved, sent and resent across a variety of distribution channels. Just maybe there is room for a player to broker that content value chain––so that the content creators can create, and the content transporters can distribute.

David Evans-NERA Economic Consulting
Success in Internet commerce often demands insight into markets in which the value of the product to one group of consumers depends on the value of a different product to another group of consumers. To take a relatively trivial example, an online dating service can’t make money from males aged 35-44 seeking females aged 25-34 who love Italian art and extreme sports unless their female counterparts have also signed up for their service. And, of course, vice-versa.

Such “two-sided” markets are not unique to the Internet, or even to what used to be called the New Economy. But a variety of factors are pushing two-sided markets to the fore. For one thing, information technology is increasingly important to the economy, and in many information-based businesses, success turns on satisfying multiple constituencies whose demand for multiple products is interdependent. And other factors, ranging from the growing importance of intangible intellectual property in business plans to globalization, seems to be accelerating the introduction of products in two-sided markets that have ‘winner-take-all” characteristics.

The complexity of operating in two-sided markets creates challenges for investors and managers, who face unfamiliar problems in identifying winning strategies. More important from the perspective of public policy, it challenges regulators to rethink the way they judge the health of competition. Indeed, it may force a basic rethinking of antitrust policy.

It’s easier to find examples of two-sided markets than to generalize about them. But three characteristics do seem common to all.  There must be distinct classes of consumers demanding complementary goods. To convince merchants to accept the Discover card (and pay fees with each transaction), Morgan Stanley must convince consumers to carry and to use the Discover card.

The value of the service to consumers on one side of the market depends on the number and qualities of consumers on the other side. Nasdaq’s services are more valuable to owners of Intel stock because the large number of potential buyers linked through the Nasdaq’s computer network increases the liquidity of the stock.
The “network economies” associated with having more participants on both sides of the market could not be realized without an intermediary. Singles in search of mates can wander around public parks on sunny afternoons looking for that perfect someone. But putting an ad in Yahoo Personals is surely a more efficient way to meet a likeminded member of the opposite sex.

Sellers maximize the difference between revenues and costs by setting prices at the point where the cost of making and marketing one more unit equals the net gain in revenue from its sale. But the price that maximizes profits in a two-sided market isn’t so easily derived, since demand for the two complementary products must be balanced. Indeed, looking at demand and cost on one side of a two-sided market alone simply can’t tell you what you need to know to create a profit-maximizing business strategy.

In fact, it may sometimes make sense to charge what amounts to a negative price to one side of the market. American Express spends more to provide ancillary services, like frequent flier miles and collision-damage waivers on car rentals, than it earns in annual fees from cardholders. But these net outlays are likely more than offset by the fees that merchants pay when the cards are used.

One could imagine a business in a two-sided market that became profitable very quickly. But, other things equal, there are good reasons to expect that the start-up costs of two-sided business will be exceptionally high. Indeed, about the best face you can put on the dot-com’s apparent indifference to profitability was the implicit assumption that, in a frontier two-sided market enterprise, the start-up period would be exceptionally long and thus profits would be a poor benchmark of success in the early years.

Note, too, that the pricing problems of firms in two-sided markets––along with the need to invest heavily in the start-up phase––may be compounded by the presence of network economies within each side. Thus new magazines often “buy circulation”––that is, offer subscriptions for little or nothing––in order to create interest among readers as well as to make the publication more attractive to advertisers. And they may offer space for little or nothing to high-profile advertisers in order to make the publication appear to be a safe bet for other advertisers.

Firms operating in two-sided markets are apt to attract the attention of competition authorities because they are likely to be more concentrated than one-sided markets. Two-sided markets invariably exhibit network effects that increase the size at which a firm will be most efficient. And daunting capital requirements typically make challenges to incumbent firms less likely. Moreover, the higher cost of establishing a foothold in a two-sided market and the higher risk that the enterprise will fail implies that a firm must expect to earn high profits if it does succeed.

Last but hardly least, no matter how competitive, costs and prices aren’t likely to track the separate sides of a two-sided market. And since it is common for one side of the market to bear much of the cost, it is also likely that, even in a very competitive market, the price charged one side of the market will be far higher than the costs directly attributable to services to that side of the market.

The issue of how firms in two-sided markets get and keep both sides on board has touched competition policy in other ways, too. Banks, which are too small to achieve network economies by issuing their own payment cards and signing up merchants to accept the cards, have formed the MasterCard and Visa associations. The payment card associations, in effect, set a floor on the fees that all banks charge to merchants accepting the cards––a practice that has, not surprisingly, caught the eye of competition authorities in many countries. Visa’s collectively set “interchange fee”––a fee paid to card-issuing banks by the bank that processed the transaction––was challenged in a private antitrust suit in 1979, but the federal courts ruled in Visa’s favor. However Australia’s central bank, which regulates the Australian card payments system, recently reached the opposite conclusion.

More generally, two-sided markets are likely to encourage the creation of novel partnership arrangements among competitors––collective behavior that might reduce competition or, as in the case of the card associations, permit an increase in competition on at least one side of the market while arguably reducing it on the other. The question for the regulator of a two-sided market is thus two-fold. Is the restraint of competition in portions of a multi-sided market justified on grounds of efficiency? If the answer is problematic is regulation in the context of interest group pressure likely to come closer to the optimum outcome?