فصل 2

Chapter 2

Winner Take All or Most

The power of a platform is the potential for rapid, nonlinear growth, especially where a company wins all or most of a market. Indeed, over the last three decades, we have seen digital platforms attain market shares of approximately 70 percent or more in relatively short periods of time. Examples include Microsoft’s Windows operating system and Office applications suite, Intel and ARM microprocessors for PCs and smartphones, Google’s Internet search technology and Android mobile operating system, and Uber’s American ride-sharing business. We also see very strong global positions at companies like Facebook in social networking, eBay in online auctions, Twitter in short social media postings (microblogs), and Airbnb in room sharing. In China, Alibaba accounts for the bulk of online shopping, and Tencent’s WeChat has a billion users and dominates messaging as well as social networking. Sina Weibo is China’s largest platform for microblogging, while Didi Chuxing has eliminated or absorbed most competitors in ride sharing.

These and many other platform companies, small and large, have benefited enormously from network effects. Yet network effects by themselves do not explain why a particular company ends up with all or most of the market or why other industries remain fragmented. In this chapter, we focus on three critical issues impacting the dynamics of platform businesses: (1) the importance of different types of network effects; (2) the impact on company performance of other factors—multi-homing (use of another platform for the same purpose at the same time) as well as niche competition and supply-side barriers to entry; and (3) how digital technologies can influence network effects and other market drivers. We especially build on work originally done on platform market dynamics by Thomas Eisenmann, Geoffrey Parker, and Marshall Van Alstyne. Network Effects: Lessons from the Telephone and Yellow Pages

Most discussions of platform markets begin with network effects. As a platform acquires more users or complementary innovations like software apps or digital content stores, positive feedback loops (i.e., network effects) emerge and can get stronger with the rising number of users or complements. The network effects make the platform increasingly valuable by attracting increasing numbers of users and complementors. Yet many people do not understand how network effects actually work. For example, even platforms with relatively strong network effects may not dominate their markets or generate much in the way of profits. Some historical examples help illustrate how network effects impact market dynamics and company performance.

We have known for more than a century that some new products and services, such as railroads and the telegraph and telephone, but also electricity, radio, and television, benefited from strong positive feedback loops with increasing returns to scale and usage. As the physical networks expanded, they attracted more users and more market participants, and opened up various monetization opportunities. For instance, American railroads in the mid-1800s started as closed systems, limited to one company and region, with incompatible track gauges. However, after companies agreed to standardize their track sizes (under U.S. government pressure), all the complying railroads benefited from a direct network effect. Thereafter, trains from one company’s network, such as in the Boston area, were physically able to connect to networks owned by companies in other areas, such as New York and Baltimore and, eventually, Chicago, San Francisco, and Seattle. A railroad that could transport people and goods from Boston to San Francisco was far more valuable than just a local system. As railroads became more useful transportation systems, they also took advantage of government land grants to build complementary businesses as well as to attract third-party investments, such as for local transportation, real estate development, banking, construction, and other services. Similarly, electricity, radio, and television were not very useful technologies without complementary innovations such as electrical appliances and programming content. As more homes adopted electricity, demand rose for more electrical appliances. As more programming content appeared for radios and then televisions, more people wanted radios and televisions.

Today, with personal computers, the Internet, and smartphones, we live in a world of platform technologies and networks connected to other platforms and networks, both physical and virtual. Yet network effects that stimulate these markets do not occur by chance. Companies and governments have to make the right strategic and policy decisions in order to drive strong network effects and make a difference in the industry. Imagine what would have happened if railroads had not standardized their track gauges. Their physical platforms would have been restricted to single companies and would have been much less useful for transportation and economic development. Or think about a world where the inventors of electric power systems had tried to control production of all the devices that would use electric power. The number of innovative products using electricity would probably have been very small. In general, network effects emerge and benefit from specific strategic decisions, or they can languish and disappear if managers and policy makers make the “wrong” decisions.

THE TELEPHONE NETWORK

A familiar historical example that demonstrates the power of network effects among multiple market participants is the telephone network. In 1876, Alexander Graham Bell patented a simple telephone device that used analog technology to replicate sounds converted to electrical impulses and relay them over copper wires. Within a few years, Bell and his financial partners organized a holding company headquartered in Boston, Massachusetts, called American Bell Telephone (renamed American Telephone & Telegraph—AT&T—in 1899). The company eventually worked with some 4,000 local and regional firms formed by independent investors to commercialize Bell’s telephone. Even today, the telephone system remains an essential communications platform, enabling billions of people to talk with each other as well as to communicate through other devices, such as fax machines. In the early years especially, the telephone benefited from powerful direct network effects. The more people who had telephones, the more other people wanted telephones. The network effects grew in strength as new telephone users inspired their friends, family, and business associates to acquire telephones. In our terminology, the initial telephone service was a one-sided platform because it targeted only individual telephone users, without making distinctions among them. Very quickly, though, the telephone business became a multisided market by identifying different segments—users who mostly made local calls versus those who needed long-distance services, and residential versus business users. Expanding usage among the different market segments became a major source of growth and profits.

AT&T in the early 1900s used the concept of network effects to justify to government regulators the decision to price the telephone service below marginal cost, which helped maintain its monopoly position. The details are more complicated than we present here, but company economists argued that universal coverage would benefit everyone on the network, and this was theoretically true. Putting telephones in nearly every home and office would lead to a nonlinear expansion of possible network connections and, in turn, an extraordinary rise in the value proposition of the telephone as a communications platform. Note in Figure 2-1 that each additional user does not increase the potential value of the network by 1. Rather—and assuming that long-distance services would eventually connect all users with telephones to each other—each additional user multiplied the potential value of the network by the number of other existing users (nodes) already connected.

Today, we refer to the network dynamic described here as Metcalfe’s law, after Robert Metcalfe, the primary inventor of the Ethernet local networking technology during the early 1970s. Metcalfe argued that the value of a communications network was the same as the number of links between its nodes. He described this relationship with a simple equation: For a network of n nodes, the value of the network is n(n – 1)/2. To illustrate, the number of potential connections for a network with two people (n = 2) is simply 1 or 2(2 – 1)/2. Five people in the network would have 10 connections. A network with 100 people would have 4,950 distinct potential connections. A network with a million users generates nearly 500 billion potential connections. What stands out is that the growth potential of a network business is not linear. (Some would say it is geometric or even exponential.) Explosive growth due to network effects enables platform businesses to expand (or decline) so quickly in scale, utility, and economic value.

Created by Derrick Coetzee (2006), public domain.

The telephone business also had to solve the chicken-or-egg problem: how to get people to use the new technology at the beginning. As long as only a few people had telephones, the positive feedback loops were weak. But even without access to Metcalfe’s law, the phone companies believed that network effects would grow in strength over time as each new user was able to connect to other users on the network. Perhaps most importantly, a subset of investors formed American Telephone & Telegraph (AT&T) in 1885 as a subsidiary of the Bell Telephone Company to enhance and monetize the network effects. The subsidiary provided lucrative “long-distance” services connecting the thousands of local telephone networks (including a few not owned by the Bell companies) to each other, especially between pairs of cities, then distant cities and rural areas, and eventually foreign countries.

In the United States, the Bell operating companies launched aggressive marketing campaigns to individuals and businesses, trying to persuade every household and organization to sign up for the new telephone service. Because telephones remained expensive, it still took twenty-five years for 10 percent of U.S. consumers and businesses to adopt them. It was another thirty-nine years before diffusion reached 40 percent of the U.S. market—much slower than the spread of television, the Internet, or smartphones. But eventually nearly every person and organization in the United States (and ultimately around the world) ended up with telephones, first landline phones and then, in recent years, cell phones.

Bolstered by very strong direct network effects, AT&T and its subsidiaries achieved close to a 100 percent market share in the United States. They maintained this dominance for more than a century. One reason was that the parent company controlled long-distance services, which made the local networks much more useful and valuable. Other market drivers and agreements with government regulators also aligned in favor of a winner-take-all outcome. The U.S. government permitted only one telephone company to exist in any one local area, and the monopoly service remained tightly regulated. For almost one hundred years, barriers to entry were extraordinarily high, and there were no differentiated or niche competitors and no true substitute technologies. Technological innovation and regulatory change eventually allowed new competitors, such as MCI (Microwave Communications Inc., founded in 1963), to use microwaves (not phone lines) to offer telephone and data services to large businesses. But there were still no mass-market alternative telephone platforms until wireless companies like Sprint and then cable TV companies and Internet-based services such as Skype gradually got into the voice communications business, most notably during the 1990s and 2000s.

FROM THE WHITE PAGES TO THE YELLOW PAGES

American Bell Telephone introduced the White Pages in 1878 along with the first telephones in order to help people find each other. This was simply a catalogue, with names in alphabetical order and divided by towns or local regions, of every individual and organization with a publicly listed telephone number. The Bell operating companies gave away the White Pages to customers for free—or really “free but not free,” because they bundled the directory cost with the telephone service. The story goes that one of the printers (R.H. Donnelley, now Dex One Corporation) realized that some users might want a separate listing of businesses and that companies might want to use the directory to advertise. In 1883 the printer temporarily ran out of white paper and switched to yellow paper for the new business listings. R.H. Donnelley formally took over printing the business telephone directory in 1886 and continued to use yellow paper to distinguish it from the White Pages. For the most part, the Yellow Pages as a business remained under the control of AT&T and its regional operating subsidiaries. The telephone company also used profits from the Yellow Pages to subsidize universal coverage, which made AT&T’s monopoly position more palatable to regulators. (Update: The U.S. government forced AT&T to divest the operating companies in 1982, after which control of the Yellow Pages business shifted to the now-independent Bell operating companies. A reconstituted AT&T regained control later on and then sold a majority stake to Cerberus Capital Management in 2012.) Clearly, the emergence of the Yellow Pages was not a random event; it involved several strategic decisions, including segmenting out business from residential telephone users, and subsidizing what we would now call the “right” side of a platform market. If the regional Bell telephone companies had simply followed their own precedent with the White Pages, they would have given away the business directories for free to all telephone users and bundled the production cost with the telephone service. However, AT&T, R.H. Donnelley, and the regional operating companies all treated the Yellow Pages as a separate business opportunity. In our terminology today, we would say that they identified commercial telephone users as a distinct set of customers and believed that this “one side” of the market would probably be willing to pay in order to connect to the “other side.” To make this idea commercially viable, the Yellow Pages had to solve two challenges, which we now know are common to all platform businesses. (See Chapter 3.) One problem was how to encourage strong network effects between different market sides—in this case, residential and business telephone users. For example, the more businesses listed in the Yellow Pages, the more users would look to the Yellow Pages to find products and services, while the more usage there was of the Yellow Pages, the more businesses would want to be listed. A second problem was the business model: Managers had to determine which side of the market to charge and how much. It was possible that telephone customers would pay extra for a business directory, but probably they would not pay very much. Although companies were accustomed to paying for advertising, usually they wanted to know how many people would view their ads. For a new offering like the Yellow Pages, viewership was a complete unknown.

How did the Yellow Pages solve these two challenges? Management decided to give away the Yellow Pages to residential customers for free and to charge companies that wanted to be listed. This decision made the Yellow Pages the primary way telephone users would search for products and services. No one could guarantee how many people would actually read a particular company’s advertisement. Nonetheless, by delivering the directory to residential phone users for free, the Yellow Pages could guarantee to every advertiser that 100 percent of the telephone customers in its locale would have a copy of the ad in their homes, whether they wanted it or not. To this day, the former Bell operating companies in the United States maintain this guarantee by continuing to deliver the Yellow Pages once per year to every customer with a landline telephone.

In our terminology, the Yellow Pages became a two-sided transaction platform. The phone companies subsidized residential customers in their search for business listings and made sure there was a wide audience for the business ads. Financially, this business model led the Yellow Pages to a spectacular run for nearly 130 years! Estimates are that the Yellow Pages franchises in the United States were “wildly profitable,” with profit margins up to 50 percent. It remained a growing business in the United States as late as 2007, worth at least $14 billion a year in sales. Of course, no business lasts forever, not even a winner-take-all platform. Today the printed Yellow Pages still exists but is mainly consulted by older people. This demographic is declining, along with the number of homes with landline telephones. Not surprisingly, R.H. Donnelley and several other companies that relied on the Yellow Pages filed for bankruptcy and reorganized or, like AT&T, sold their interest in the business. The Yellow Pages has survived as an online digital service and as a paper directory. Although this is not likely to continue for too many more years, nearly 70 percent of U.S. households continue to receive the Yellow Pages. Most local businesses also continue to pay hundreds and even tens of thousands of dollars per year (New York City rates were the highest) to advertise in the Yellow Pages. THE YELLOW PAGES COMPARED TO DIGITAL PLATFORMS

If the Yellow Pages story sounds familiar to a modern reader, well, it should. We have seen similar business models many times and in different forms: A company gives away part of a product or service for free or at low cost to stimulate adoption, and then charges for another essential part of the product or charges another market side. People and organizations paid for the telephone service and received the directories for free, and companies paid for advertisements in the Yellow Pages. As for more modern examples, we know that Microsoft licensed DOS and Windows at low prices to PC makers and attracted applications producers with free software development kits. Adobe made its Acrobat Reader free for individual users and then built up a multibillion-dollar business around enterprise server software and editing tools. Google, Facebook, and Twitter built up their user numbers through free access and then sold advertising. Alibaba built its online marketplaces by making listings free but charged for preferential placements as well as transaction fees on sales in some of its marketplaces (such as Tmall).

Over the past two decades, Google Search and other search engines have gradually replaced the printed Yellow Pages, but we can still compare the business models. Both the Yellow Pages and Google Search provided free-of-charge “windows” into information networks (one on paper and the other digital) and made it possible for two market sides to find each other. Both sold advertising to the one side that most wanted to reach the other side. The Yellow Pages was really a complement to the once tightly regulated telephone system in the United States. Google Search was really a complement to the global Internet. It took the dismantling of regulation to undermine AT&T’s telephone monopoly as well as digital competition to undermine the 100 percent market share of the Yellow Pages business in the United States. AT&T and the operating companies controlled this business, sanctioned by government regulators. The comparison is not exact, but Google Search attained a 90 percent market share in most parts of the world without government support. In fact, in some countries, such as China and Russia, the governments limited Google’s operations. In these markets, Google Search had dramatically lower shares compared to local firms, which illustrates the power of government regulation.

Other Market Drivers: Multi-homing, Differentiation/Niches, and Entry Barriers Strong network effects are a powerful driver toward a winner-take-all or winner-take-most market outcome. But we have pointed out that network effects alone are not enough because other factors, in addition to government regulation, also contribute to the market shares gained by particular firms. With instant messaging, for example, there are very strong same-side network effects—just like the telephone. Yet most messaging platforms, ranging from ICQ and Yahoo on personal computers to WhatsApp and BlackBerry Messenger on smartphones, never made money. One exception seems to be WeChat in China, which has expanded from a messaging (transaction) platform into an innovation platform such as for video games and payment services, with several ways to make money, albeit with protection against global competition from the government. (We will say more about WeChat later.) But why, in general, was it so hard to tip and monetize the market for instant messaging? Because a market is unlikely to coalesce around a particular platform if users can easily use multiple platforms at the same time for the same purpose; if competitors have differentiated or niche platforms that divert users with unique features; or if new firms can easily enter the same market and compete with better services or lower prices. All these conditions generally apply in the instant messaging market, as well as many other industries, conventional and digital. So let’s examine in more detail three other market drivers that can reduce or enhance the power of network effects.

MULTI-HOMING

No company in a conventional or platform business likes to see its customers use products and services from competitors. In Michael Porter’s framework, the more competitors there are, the more intensive the rivalries. Intense rivalries often lead to price competition that can reduce profits for everyone. Platform owners want their customers to stick to one platform, whether it is for computing, cell phone devices and services, online shopping, room renting, or ride sharing.

In contrast to conventional businesses, many platforms (e.g., the Yellow Pages, Google Search, Bing Search, the Android operating system, Facebook, or WeChat) did not charge users directly. For these types of platform businesses, there was no immediate reduction in sales if users multi-homed because the platforms did not directly sell a product or service. Instead, multi-homing weakened network effects, which these platforms depended on to attract other market sides, such as advertisers or producers of complementary innovations. So we can see that multi-homing by users can inhibit a platform even with strong same-side (direct) network effects from fully monetizing cross-side (indirect) network effects.

Think about Twitter, a classic example of strong direct network effects. Popular tweeters attracted followers and encouraged more tweeters and more followers. Similar to Google, Twitter monetized its free services by selling advertisements. However, until very recently, Twitter made little or no profit because of high operating costs, the expense of new customer acquisition, and relatively low advertising revenue. Part of Twitter’s problem was that many tweeters multi-homed. Twitter users spent time (usually more time) on Facebook, Instagram, Snapchat, or WhatsApp to communicate about personal or detailed matters, such as vacation plans and tastes in music or movies. Consequently, the time and attention as well as the personal details of Twitter users, along with their purchasing power for third-party advertisements, were divided among multiple platforms—despite Twitter’s strong direct network effects.

The key point is that multi-homing impacts network effects and, indirectly, impacts a platform’s potential revenues and profits. Therefore, how to limit multi-homing is an important goal for all platform companies. The solutions are not always obvious. Some innovation platforms build proprietary standards (e.g., Microsoft Windows or Apple iOS), while others tie complementary services to the platform (e.g., Google Android and the Google Play Store for smartphone apps). Some transaction platforms create loyalty programs (e.g., Expedia), following the lead of airlines and credit cards. But some platform companies also make mistakes or find themselves making choices they would have preferred to avoid. In particular, whereas low prices on the “right” side of the market can attract users, low prices on the “wrong” side can encourage multi-homing and weaken cross-side network effects.

Video games are a good example illustrating the risks of competing with an aggressive subsidy to one side of the market without overcoming potentially negative consequences from multi-homing. In the early and mid-2000s, Microsoft and Nintendo decided to sell their new game consoles (Xbox and Wii, respectively) at near cost, that is, for a few hundred dollars. Sony responded by keeping prices relatively low on its PlayStation, though generally still higher than the other consoles. All three companies decided to subsidize the user base and make most or all of their profits by charging game developers high license fees and royalties to write games for their platforms. The console makers hoped that a large and growing base of users, attracted by low console prices, would exert a strong pull on the complementor side of the market, the game software producers, and eventually produce a steady stream of revenues and profits. Microsoft, Nintendo, and Sony developed some games themselves and invested in studios to seed the market, but they also saw the need to foster a vibrant independent ecosystem of game developers. In the video game market, each console was technologically incompatible, so Sony, Microsoft, and Nintendo were able to line up some exclusive content for their platforms. Only the biggest developers had the resources to write software for more than one platform.

But the strategy adopted by the platform leaders resulted in an unintended consequence: Low console prices encouraged multi-homing. Serious video gamers (mostly teenagers and young male adults) generally bought more than one console, especially those that had compelling games. These “hit” software products also varied with each new generation of the hardware. As a result, we have not seen one company take all or most of the video game market and sustain that lead over multiple product generations, as Microsoft has done with Windows for personal computers or like Google has done with Android for smartphones. Market shares have varied from console generation to generation, depending on which company had the most compelling new video games or differentiated hardware features.

DIFFERENTIATION AND NICHE COMPETITION

All companies worry about competitors that make their products and services stand out through better quality or serving the special needs of particular types of customers. Even platforms that do not directly sell a product or service need to worry about differentiated and niche competitors. Similar to the challenges created by multi-homing, a fragmented market with niche players reduces network effects and the likelihood of a winner-take-all outcome. The more homogeneous the market, the higher the likelihood that strong network effects will attract the vast majority of users, which could drive the market to tip toward one platform. Think about smartphones: On the surface, one might have expected the market to tip to the player with the early lead in market share, much like the PC market tipped to computers running DOS in the 1980s and then Windows in the 1990s. Apple was the early leader in the modern, touch screen smartphone market after introducing the iPhone in June 2007. However, devices running Google’s Android eventually dominated the market.

One particularly important decision Google made to help tip the market dates back to November 2007. This is when Google mobilized handset makers that wanted to compete with Apple by forming the Open Handset Alliance. Network operators and software developers also joined the coalition and agreed to promote “open standards” (i.e., technologies that multiple companies could freely license and use). Most importantly, members (i.e., Android licensees) could freely use the software as long as they agreed not to create incompatible or “forked” versions. Google wanted to maintain control over the operating system and make sure that handset makers and software developers continued to use Google services, such as its search, browser, and mapping-location technology, which Google used to sell advertisements. Some forked versions of Android still appeared. However, in general, Google’s strategy worked because Apple would not license its technology to anyone for any price. Android provided a compelling alternative (free, good enough, and open to improvement). Apple failed to keep a majority of the smartphone market, which it briefly achieved after introducing the iPhone. Nonetheless, Apple’s highly differentiated product and growing ecosystem of app developers allowed it to keep high-end customers. Several analysts even estimated that Apple earned over 90 percent of handset industry profits with less than 18 percent market share during 2015–2017. Apple’s ability to differentiate the iPhone as a product and a platform enabled it to charge premium prices (although this may change in the future) and prevented Google from taking more than 80 percent of the total smartphone market.

Another illuminating example is the market for freelance labor, dominated in 2018 by Upwork, a classic transaction platform created by the merger of Elance and oDesk. Similar to many transaction platforms, Upwork experienced powerful cross-side network effects: The more freelancers who joined the platform, the more value companies realized from using Upwork; and the more corporations looking for freelance labor, the more value seen by freelancers around the world that seek employment on Upwork. Indeed, by early 2018, Upwork announced that $4 billion of cumulative work had been completed through the platform and its annual run rate of gross services volume had reached $1.5 billion. Upwork reported that 28 percent of Fortune 500 companies posted jobs on the platform in 2017, and the company’s platform had 5 million corporate clients and 12 million freelancers.

However, CEO Stephane Kasriel summarized the challenge for Upwork, estimating that there were at least “500 competitors in our space, most of them focusing on small niches.” Despite the obvious power of network effects and Upwork’s strong brand and growing platform, niche companies thrived by focusing on specific industries, specific job types, and specific geographies (local as well as global). Upwork’s horizontally designed platform was competing against vertically specialized players who had more local expertise and could siphon off both corporate clients and specialized freelancers. Despite two decades of operations and strong growth, continued fragmentation prevented Upwork from tipping the market or even making money. When the company filed its S-1 in late 2018, it was still not making an annual profit. (Nonetheless, the company went public on October 3, 2018, with a 40 percent surge in the stock price on the first day.)

BARRIERS TO ENTRY

All companies share a desire to limit new competitors from entering their markets. In general, if switching costs are low and entry is easy, then markets are unlikely to be highly profitable. Low entry barriers encourage more competition, which generally leads to lower prices and lower profits for everyone. In the platform world, most companies focus strategically on the demand side of the market: how to get more users (customers). But when traditional barriers to entry are low, even in markets where companies feel protected because of their strong network effects, new entrants can still enter the business on the supply side and fragment the user base, preventing a market from tipping toward one big winner.

The unique dilemma for many platform businesses is that the initial cost of market entry can be very low because of advances in digital technology, which we will discuss in more detail below. In the world of lean start-ups, the amount of capital required to develop, produce, and distribute a new product or service, or even a new platform, is a fraction of what it cost ten or twenty years ago. In the gig economy, it has been especially easy to start new transaction platforms like handyman services (e.g., Handy or TaskRabbit). In Chapter 4, we discuss how twenty-nine companies entered on-demand services, even though few survived. Similarly, dozens of companies have entered the market for online communities, web portals, and B2B marketplaces, mostly because the entry costs were relatively low.

In markets where barriers to entry are high, we see a different pattern: much more industry concentration, with a higher probability of the market tipping toward one or a small number of firms. In capital-intensive businesses, like developing new cloud services and related innovation platforms, there are a relatively small number of dominant firms (e.g., mainly Amazon, Microsoft, and Google, followed by IBM and Alibaba). Similarly, costs can be very high for subsidy-intensive businesses like automobile ride sharing (e.g., Uber). In addition, when companies such as Qualcomm have been able to protect their market positions with patents, unique technical know-how, government regulations, and other barriers to entry, markets have had a higher likelihood of tipping toward one platform.

There are also unique barriers to entry in platform markets that we rarely see in the conventional world. First, network effects create barriers in the form of an existing stock of complements around a particular platform. When a platform has millions of applications that run on—and only on—Android, iOS, Windows, Amazon Web Services, Facebook, or WeChat, they increase the barriers to switching to a new or competing platform. Second, and closely related, is that new entrants are often challenged by the difficulty of replicating a platform ecosystem. Successful platform companies have established virtual armies of complementors, such as the thousands of software developers who have already joined the Android, iPhone, Facebook, or WeChat developer networks, or the millions of people with rooms to rent or cars to drive who have already registered with Airbnb, Uber, Lyft, and Didi Chuxing. As the number of complementors grows, it becomes increasingly difficult for a new firm to enter late and build a competing ecosystem for the supply side of the platform. Third, networks themselves create complex switching costs. When the value of the platform depends on the number of complements and users who are directly connected, then it can be extremely difficult or expensive to switch. For example, if people want to stop using LinkedIn and move to a new professional network, they must convince their professional contacts to move with them, otherwise the new platform will have relatively little value.

Digital Technologies: Impact on Platform Market Drivers

To summarize: At any given point in time, the likelihood of a winner-take-all-or-most outcome in a platform business will depend on the strength of network effects, the difficulty of multi-homing, lack of opportunities for competitor differentiation and niche competition, and the strength of entry barriers. At the same time, though, we live in a world where digital technologies are rapidly changing market dynamics. Moore’s law drove the most fundamental market transformations, with a doubling of computer processing power every eighteen to twenty-four months from the 1960s until recent years. Once the personal computer emerged in the late 1970s and 1980s, a new generation of computing platforms appeared for basic software and applications. With essentially zero marginal costs for digital goods, the economics of platform businesses would change forever. And once the World Wide Web became a mass-market phenomenon after the mid-1990s, managers and entrepreneurs had their first opportunity to create truly ubiquitous global platforms. With new platforms came new opportunities to innovate and exercise different forms of economic, social, and political power, much of which we now associate with digital technologies.

During the last decade, the combination of mobile and cloud technologies, artificial intelligence and machine learning, as well as big data, has accelerated the diffusion and refinement of digital platforms. Advances in technology also have brought several formerly separate markets together. This “digital convergence” has been going on since the mid-1990s, at least. It helps explain why a single company—Apple—passed a trillion dollars in market value in August 2018, before falling back. The smartphone, which accounted for 60 percent to 70 percent or more of Apple’s sales over the past decade, was at the same time a phone, a computer, a digital media player, a digital camera, a digital recorder, a personal digital assistant, a video game machine, and a handheld television, among other things. Apple’s value reflected the integration of products and services in all these markets, including rapidly increasing revenues and profits from its digital content store (iTunes) and transaction platform (the App Store). Mastery of digital technology and a hybrid platform strategy also helps explain why Apple, Amazon, Google, Microsoft, Facebook, Alibaba, and Tencent were among the most valuable companies in the world.

In general, digital technologies can either help or hurt incumbent platform leaders, and they can either help or hurt new platform entrants. As a result, the impact on platform businesses is complex and requires a separate look at how technological innovation can affect the four fundamental drivers of platform market dynamics.

IMPACT ON NETWORK EFFECTS

What comes to mind first is that digital technologies should strengthen network effects. In fact, if one compares the potential for network effects in today’s world with pre-digital times, the differences are stark. Think about how long it took to build out a railroad in the United States or to get broad adoption of the telephone and Yellow Pages or even credit cards. Numerous studies have shown that adoption rates have accelerated with new technologies, and digital platforms are no exception. Digital technology also enables platforms to connect faster and to link more people and organizations with each other than ever before.

For example, it was possible to use Facebook to connect with more than 2 million websites between 2008 and 2010, including 90 percent of the top 1,000 sites then on the Internet. At one point, that number was growing by about 10,000 websites a day. More importantly, about one-third of Facebook’s users interacted with the social network through third-party sites every month. As early as 2012, some 9 million applications and websites already operated within or were accessible through Facebook. Since then, a growing portion of Internet activity has involved Facebook even without happening directly on the Facebook platform.

New software programs for analyzing massive amounts of data on user behavior can further enhance the power of network effects. However, the general principle is that, as a digital platform collects more data and then applies machine learning and other artificial intelligence algorithms, the platform can become “smarter.” For example, it can create more targeted advertisements or better search results and recommendations for additional purchases. Furthermore, as users contribute their own data in the form of content or ratings (as they do with Google, Amazon, eBay, Facebook, Twitter, Instagram, Snapchat, WeChat, Expedia, TripAdvisor, Uber, Airbnb, and many other platforms), data and analytics can also help improve the product or service. The big and the wealthy can get bigger and wealthier because they have more data and more money to invest in technology and marketing. This dynamic, driven by network effects, is surely part of why platform businesses have become so attractive to managers, entrepreneurs, and investors.

Clever firms also can use digital technologies to strengthen network effects in completely new arenas. Take Waze, the Israeli road navigation app created in 2008 that Google purchased in 2013. Waze users not only consumed data but they also generated data, constantly and in real time. This app originated as a one-sided platform analyzing driver information. Later, it started to sell advertisements. Then it began to provide traffic information to TV and radio broadcasters as well as city traffic authorities for free, thereby increasing brand recognition and attracting more users and advertisers. Waze even added a social media element so that registered users in close proximity could contact each other and share information as they were driving. In addition, Google integrated information from Waze into Google Maps, to make its traffic information more accurate. As users drove with the Waze app open, they automatically generated data about their location and speed, while users added their own reports on accidents and other slowdowns. Waze computers captured and analyzed these data (initially with the help of volunteer map editors) and made suggestions for alternative routes in case of heavy traffic or accidents. The service continually improved, depending on how much users consumed and contributed to the service. (It is also true that Waze and similar navigation apps from Apple, Google, and other companies often directed drivers to the same alternative routes and sometimes created new traffic jams.)

IMPACT ON MULTI-HOMING

While digital technologies have been creating more opportunities across the board for stronger network effects, those same technologies can facilitate as well as reduce multi-homing. How can this be? It depends on how well companies plan and execute their digital strategies. Especially with modern transaction platforms, the tangible costs to multi-homing seem trivial. In the old days, it was expensive and difficult for most users to own both a Windows PC and a Macintosh. Most users chose one platform and were restricted to applications available on that platform. Nowadays, applications are usually available on both types of personal computers and many more applications are available as web applications, accessible from different types of devices and platforms. For example, it costs nothing to do a general search on Google, compare airfares on Kayak, or seek travel advice on TripAdvisor or Expedia. There is no need to buy a particular computer or smartphone. All users need today is a device with access to the Internet.

In a purely digital world, we expect that consumers will have low cost or free alternatives widely available for nearly every activity and that they will be multi-homing as a matter of course. Google executives have made this argument, citing the ease of multi-homing to defray criticism of their dominant position in Internet search, since competitors are merely “one click away.” As we discussed earlier with the Facebook example, Mark Zuckerberg may have created the world’s largest social network. However, it is easy for Facebook users to spend time on Twitter, LinkedIn, Snapchat, Pinterest, and other platforms, even if they may not take the trouble to adopt another social network for most of their activities. To control some of the revenues associated with multi-homing was at least one reason why Zuckerberg acquired Instagram in 2012 for $1 billion (considered a large sum at the time) and later used it to compete with Snapchat. Zuckerberg’s much more costly purchase of WhatsApp in 2014 for what amounted to $19 billion plus another $3 billion in Facebook stock was another defensive move against multi-homing for messaging as well as an offensive move to acquire more users and a potentially new revenue source. At the same time, clever use of data and sophisticated AI tools can discourage multi-homing because of better services. Waze and other digital platforms can analyze customer behavior and then refine platform functions down to the lowest technical levels, ranging from the design of user menus to the display of content and recommendations. Unlike the telephone or the Yellow Pages, sophisticated data analysis tools combined with the emergence of big data can help a digital platform provide more effective automated responses to user requests, more targeted searches and ads, better suggestions, or even more attractive prices—every time a customer uses the system to communicate or to look for something to buy. The result is that some digital platforms use their technological expertise and scale to offer such compelling services and prices that consumers often do not bother to multi-home. Almost 50 percent of American online shoppers simply start and end on Amazon; there is a similar pattern in China on Taobao. Or users do not bother to look at multiple search engines; they simply start and end on Google or, in China, on Baidu.

Digital technologies also discourage multi-homing as well as switching in the B2B world. Once a business integrates application stacks and platforms like SAP, Salesforce, Amazon Web Services, or Microsoft Azure into the fabric of their organizations, the cost of adopting a competing platform often goes up, even with new portability standards for cloud data and applications. Similar to the dynamics of enterprises adopting IBM mainframes in the 1960s or Windows personal computers in the 1990s, once an organization has committed to a particular technology, it generally spends significant resources to train its employees on that platform and builds specialized applications and interfaces or adds customized features.

IMPACT ON DIFFERENTIATION AND NICHE COMPETITION

There are many books and articles already written on how firms can use digital technology to make their products and services stand out or better serve niche customers. The challenge is that the digital revolution has made it easier to create new niches as well as easier for competitors to copy them. There are many variations in how platform companies have used their technological expertise to differentiate themselves or pursue markets. A few cases illustrate what is possible.

One well-known example is Snapchat, the widely used messaging app for millennials. Snapchat’s cofounder, Evan Spiegel, had a simple idea: Young smartphone users hated the idea that their messages lasted forever. Whether it was a parent checking your phone, or your girlfriend or boyfriend seeing old messages, millennials wanted some privacy. Moreover, millennials liked their chat messages to tell a story. With digital technology, solving this problem was not technically complicated. Spiegel was able to quickly create a new messaging app in a crowded space with features that allowed messages to disappear after a specified time. As Spiegel wrote in his first company blog in 2012, “Snapchat isn’t about capturing the traditional Kodak moment.” He wanted to let millennials avoid the stresses caused by the longevity of personal information on Facebook and other social media. The result was a high degree of differentiation from Facebook, Twitter, and other messaging platforms, and a viral explosion of users. By 2018, there were roughly 190 million daily active Snapchat users.

As for facilitating niche competition, we only need to look at the wide range of specialized online shopping sites to see how digital technology prevented the giant online stores and marketplaces at Amazon (globally), Alibaba (in China), and Rakuten (in Japan) from getting close to 100 percent of the Internet shopping business even in their home countries. In the United States, Walmart, Target, and every other major retailer created their own online sites and made acquisitions to compete with Amazon. (See Chapter 5.) In addition, Amazon had to contend with a growing number of niche marketplaces and digital stores. For example, Star 360 (starthreesixty.com) sold shoes for men and women, representing nearly every major brand. Koovs (koovs.com), Lifestyle (lifestylestores.com), and PrettyLittleThing (prettylittlething.com) were online fashion portals representing major brands and offering large discounts as well as dozens of new fashion items every week. Bluemercury (bluemercury.com) specialized in beauty products. Horchow (horchow.com), owned by the Neiman Marcus Group, sold furniture. Etsy (etsy.com) dominated the marketplace for handmade items ranging from clothing to arts and crafts. Winemonger (winemonger.com) sold wine, which users could search for by country and type. The list goes on and on for specialized retail platforms and digital stores.

The flip side of digital technology accelerating opportunities for differentiation and niche competition was that it could be equally easy for incumbents to copy. Mark Zuckerberg almost immediately recognized Snapchat as a potential threat to Facebook. Given Facebook’s size and scale, he logically tried to buy Snap for $3 billion in cash before the company went public. When Spiegel turned him down, Zuckerberg ordered Instagram to copy Snapchat’s most compelling features and turn Instagram into a Snapchat killer. Especially after introducing Instagram Stories, Instagram zoomed past Snapchat, with over 700 million users. Although Snapchat was hardly dead (its market value had fallen dramatically but was still close to $6 billion in late 2018), Facebook’s attack had taken a serious toll. By mid-2018, Snapchat’s user base had begun shrinking for the first time in its history. Both new entrants and established firms can also utilize their expertise in digital technology to rapidly copy niche strategies and integrate acquisitions. When Amazon has observed (such as by analyzing transaction data on Amazon Marketplace) a new category that was growing fast, it frequently replicated its competitors’ online platforms or just bought them outright. Amazon followed this strategy when it bought the online shoe retailer Zappos in 2009 for $1.2 billion and the video game streaming company Twitch in 2014 for $970 million, as well as the Middle East’s biggest online retailing site, Souq, in 2017 for $650 million. Amazon also combined Whole Foods, which it acquired in 2017 for $13.7 billion, with AmazonFresh and its online grocery business. Similarly, when eBay pioneered online auctions, many firms, including Yahoo and Amazon, entered the market with new features or targeted vertical markets. In a non-digital world, replicating those features and verticals might have been too expensive or taken too much time to be competitive. But eBay was able to quickly copy the best new features (e.g., insurance) and introduce its own vertical sites (e.g., eBay Motors).

IMPACT ON BARRIERS TO ENTRY

Digital technologies, especially when they strengthen or accelerate network effects, have both lowered as well as raised entry barriers into many industries. How can this be? With cloud computing and nearly ubiquitous Internet access, establishing new digital platforms has become easier than ever. Twenty years ago, firms had to build their own data centers and invest heavily in computing power to begin a digital business. With the advent of the cloud, new businesses can get started practically overnight. The cost of building a new software company or transaction platform focused on the mass market or particular segments, and launching it on Amazon Web Services, Microsoft Azure, or other cloud services platforms, has dropped dramatically. As more businesses are built in the cloud, it has become easier to connect more consumers to consumers, more businesses to other businesses, and more businesses to consumers. The rising number of new transaction platforms among the billion-dollar unicorns, especially those competing within the gig economy, may also help explain why so many new platforms have been unprofitable as businesses.

At the same time, digital innovations have enabled new scale and scope economies, or enabled entry for some and raised entry barriers for others, in ways unimaginable in the pre-digital era. Think about how Amazon, founded by Jeff Bezos in 1994, expanded from being an online store selling books to an online store selling nearly everything, from electronics products to groceries, and with same-day delivery for some products. Even in the early days, Amazon used digital technology to promote online store sales, building a recommendation engine and collecting user evaluations. One estimate is that 40 percent of Amazon’s sales today come through its recommendation engine. Then, in the late 1990s, Bezos added the global Amazon Marketplace—what we have called a transaction platform—linking buyers and third-party sellers. Amazon combined the marketplace with its own online store and other fulfillment services, such as billing and shipping, in addition to a massive network of physical warehouses. Competing with Amazon has become a scale game, in which even Walmart, the largest company in the world by revenues, has struggled.

But “platformizing” a market with digital technology does not necessarily change the fundamentals of a business or make common sense and domain knowledge obsolete. For example, simply creating a digital platform to enter the grocery business with online ordering does not make groceries as profitable as selling digital goods or make irrelevant a deep understanding of how to handle perishable goods. An online vendor like Amazon must still deliver groceries in the real world and understand how different supply chains work. If groceries are going to be profitable for Amazon as an online business, then it will probably be because the company can link different businesses and assets in unique ways and achieve economies of scale and scope that other firms cannot. Or perhaps the secret will actually be the physical network of warehouses and delivery vehicles that Amazon invests in to complement its digital platforms.40 The leading Chinese platform companies also have used their digital expertise as well as knowledge of local markets and institutions to make it difficult for global platforms like Amazon to compete in China. Alibaba and Tencent started with “asset-light” platform business models: Their early operations were purely digital with modest capital requirements. But, similar to Amazon in the United States though at a somewhat lesser magnitude, Alibaba and Tencent have turned e-commerce in China into a scale game. In 2018, Taobao Marketplace controlled about 60 percent of China’s B2C e-commerce. Alibaba also used its size and technological resources to enter cloud computing, payment services, and other related businesses, making it very expensive for Amazon or Walmart to compete effectively with Alibaba in its home market. The same was true of Tencent, which dominated Chinese messaging and social media with WeChat. Tencent exploited its size, customer base, and digital skills to add social media applications as well as expand into online payments, small-business credit and investment services, digital entertainment, and video games, among other areas.

Key Takeaways for Managers and Entrepreneurs

In this chapter, we discussed the fundamental drivers of platform markets and the dynamics of a winner-take-all-or-most outcome. The most important factors are network effects among users and different market actors, such as between end users and advertisers or between end users and producers of complementary innovations. But we also pointed out how multi-homing and competition from differentiated or niche competitors can reduce network effects and monetization opportunities. Low entry barriers as well can lead to increases in the number of competitors and equally weaken network efforts and profitability. So when it comes to understanding the drivers and dynamics of platform markets, what are the key takeaways for managers and entrepreneurs?

First, it is important to understand what network effects really are, where they come from, and what impact they have on competitive advantage. We have defined them as self-reinforcing feedback loops that, ultimately, create platform value, directly or indirectly. The biggest business challenges for platforms are to nurture network effects and then translate the momentum and value created into a steady and growing stream of revenues and profits. More specifically, same-side network effects come directly from connecting users with other users. Cross-side network effects come from connecting different market actors to users. Platforms make money by facilitating these connections and associated innovations. In both cases, network effects are not simply abstractions. They derive from specific strategic decisions and investments, and avoidance of actions that can depress network effects, such as pricing too high on the “wrong” side of the market. Imagine what would have happened if Facebook had charged users to access the social network, like social clubs in the physical world? It would probably be a small, niche business today, or Facebook might have failed completely. It is also essential to realize that network effects can weaken or collapse as technology and market dynamics change, sometimes with little advance warning. This occurred when Apple introduced the hugely disruptive iPhone in 2007 and, in effect, destroyed the nascent smartphone businesses of Palm, Nokia, BlackBerry, Microsoft, and other companies.

Second, it is no accident that the world’s most valuable public companies are platform businesses born after the emergence of the personal computer, the Internet, and the smartphone. But we also saw in this chapter that enabling and sustaining platform dominance requires more than just network effects. Successful platforms find ways to encourage users and third-party complementors to adopt their platforms and innovate on top of them. They make it difficult for users and complementors to use or switch to competing platforms. They may adopt other measures, such as forming coalitions or using subsidies to help tip a market toward their platform when multiple platforms compete. Remember, though, we learned from Apple that a platform does not have to get a majority of industry revenues to get a majority of industry profits as long as it focuses on the most profitable customers.

Third, platforms enable “asset-light” business models by connecting different market actors and leveraging network effects. As we discuss in Chapter 3, an innovation platform can facilitate and then take advantage of new products and services built by third-party firms that continuously make the core product or service—the platform—more valuable. A transaction platform can facilitate and then take advantage of interactions among market participants that would not otherwise occur, and get one market side to pay for access to another side or to supply critical assets. Think about how difficult and expensive it would have been if Microsoft, Apple, Google, Facebook, or WeChat had tried to hire all the engineers that built the millions of applications now available for their innovation platforms. Think about how expensive it would have been if Airbnb had tried to buy all the homes and apartment buildings its users accessed, or if Uber, Lyft, or Didi had tried to buy all the vehicles its drivers use.

In short, we learn from looking at market fundamentals that a platform business must compete in multiple dimensions. Some of these dimensions (like reducing the ability of other firms to exploit differentiation or niches and erecting entry barriers) are the same as competition in conventional markets. Other dimensions (like generating network effects or limiting multi-homing) are distinctive to platform businesses. Ultimately, however, platform companies must offer a compelling product or service that is superior to what the competition offers, whether the other firms are digital or conventional businesses. A successful platform company must be able to protect its competitive advantage in order to retain customers on the demand side as well as attract workers and asset providers on the supply side. If a platform cannot do these things better than the competition, then the company will lose money, like any other business.

Diving more deeply into the differences between innovation and transaction platforms, as well as the advantages of a hybrid strategy, is the subject of the next chapter.