فصل 4

Chapter 4

Common Mistakes

Mispricing, Mistrust, Mistiming—and Hubris

The inevitable result of pundits proclaiming a digital revolution has been a mad rush into the platform business. Similar to the first dot.com boom of the 1990s, there has been a predictable drive to become the first and the largest new platform in every space. It looks like a land grab, where companies feel they have to be the first mover to secure a new territory, exploit network effects, and raise barriers to entry for future players. Uber’s frenetic efforts to conquer every city in the world at breakneck speeds and Airbnb’s desire to create the platform for sharing rooms all over the world are the two most obvious recent examples.

The problem is that being the first mover has never been a guarantee of success in platforms or non-platform businesses. Some early movers, such as Amazon in e-commerce or Apple in a new generation of smartphones, have translated into powerful positions. However, the platform world has been littered by the failure of first movers in almost every category. Sidecar—not Uber—pioneered the ride-sharing market; VRBO and numerous vacation rental platforms—not Airbnb—pioneered the rental room market; Friendster and MySpace—not Facebook—pioneered and dramatically expanded social networking.

The power behind the myth of the first mover is FOMO: the fear of missing out. In markets where someone solves the chicken-or-egg problem and network effects kick in (like the Yellow Pages or Google Search), the consequences of failure can be devastating: Losers get locked out of markets indefinitely. But more often than not, first movers stumble. Friendster is a good example. It built a huge social network before Mark Zuckerberg, but Friendster made poor technology choices, which did not scale. Frustration with Friendster, which could take up to twelve seconds to load a single web page, left open the window for Facebook to explode onto the scene. eBay, which dominated e-commerce in the United States, was also the first mover in China. While eBay quickly grabbed a dominant share of the Chinese market, Alibaba overwhelmed it within a few years. Netscape captured 80 percent of the early browser market in the mid-1990s, yet it was crushed by Microsoft; then Microsoft had over 90 percent of the market, and ended up losing leadership in recent years to Firefox and then Google’s Chrome.

Patterns in Platform Failures

To understand why and how platforms fail, we tried to identify as many failed American platforms as possible over the last twenty years. We reviewed the annual reports and proxy statements of the innovation and transaction platform companies that we analyzed in our database (see Chapter 1 and Data Appendix) and made a list of the competing platform companies they mentioned. The 209 failures that we found, while by no means exhaustive, allow us to make some general observations about why platforms struggle.

First, as shown in Appendix tables 4-1 and 4-2, the most obvious pattern is the predominance of failed transaction platforms, representing nearly 85 percent of the sample. Transaction platforms have a somewhat shorter life as well, averaging 4.6 years, compared to 7.4 and 5 years for the hybrid and innovation platforms we identified, respectively. Not surprisingly, many platforms, like Friendster, failed because they were weak platforms: Their technologies were outdated or their user interfaces were complicated and hard to use. Especially for transaction platforms, the barriers to entry for starting a marketplace have been low, and these platforms often failed because they could never get enough market participants on one or more sides of the platform to achieve positive network effects and take off. Many gig-economy platforms collapsed within two to three years because they did not have enough users and ran out of funding. One of the challenges for those firms in the local delivery and services space or in ride hailing was that the network effects were local, but the only way to rapidly get to scale and build brand recognition was to expand geographically. To do so required attracting more investment and having deep enough pockets to go potentially a long time before reaching profitability and positive cash flow.

Given the need for deep pockets, it should not be surprising that stand-alone firms tended to have shorter lives than those that were acquired or launched as part of a larger firm or consortium of firms. Overcoming chicken-or-egg problems was apparently much more challenging for stand-alone firms, which had an average duration of only 3.7 years. Acquired firms, which generally had stronger balance sheets, were capable of fighting longer (average 7.4 years), while firms that were part of larger entities averaged 4.9 years.

Consolidation was also a common pattern among platforms in the same space. Platforms frequently disappeared because they merged. This was especially true in certain categories. For example, a wave of acquisitions reduced the large number of web portals and search engines that emerged in the 1990s down to a few players by the mid-2000s. More recently, peer-to-peer car-sharing platforms saw consolidation as larger rivals acquired two major platforms. In addition, there was considerable consolidation among business-to-business marketplaces that emerged in the 1990s dedicated to specific verticals—such as exchanges for airplane parts, medical supplies, or chemicals.

In a relatively small number of platform spaces, failure was a function of a genuine winner-take-all or winner-take-most outcome for a competitor, even without an assist from government regulation. Mobile operating systems and social networking applications were prime examples, which we discuss later in this chapter. No third mobile operating system gained significant share since the market tipped to iOS and Android, despite the fact that competitors like BlackBerry, Windows, and Symbian entered the market first and had significant resources at their disposal. Similarly, in social networking, Facebook displaced earlier rivals like MySpace, Friendster, and GeoCities as the market tipped.

Social networking illustrated another pattern: Failing platforms sometimes sought out a niche in order to survive. The results, however, were mixed. For example, Ello pivoted from a general social network to a niche social network as a space for creative collaboration where artists could display their work and receive feedback from other artists. Though iVillage failed, it survived for a long time (eighteen years) with a focus on women, admittedly a very large niche. Disney’s Club Penguin survived for twelve years until 2017 with a focus on kids, while Path tried a geographic focus on Asia. On the other hand, Kinly, a private social network for families, never took off and folded after two years.

Failure as a platform did not necessarily mean failure as a company. A number of online marketplaces (especially B2B) that failed to generate sufficient trading volume pivoted into adjacent businesses. One example is seafood.com, which launched in 1996 as a news service for the seafood industry. In 1999, it added both wholesale and retail marketplaces to the site, while continuing to provide industry news, including a subscription news service. In 2012, it dropped the marketplace and reverted to the news-only site, supported both by advertising and subscription revenue; as of 2018, it continued to operate as an industry news service. Another example was Exostar, launched in 2000 by Raytheon, Boeing, Lockheed Martin, and BAE Systems as an independent marketplace for the aerospace and defense industry. Over time, it pivoted away from its marketplace function and positioned itself to foster supply-chain collaboration in the industry, and later for a variety of industries. By 2018, Exostar had shifted its business model to one of “helping organizations in highly-regulated industries mitigate risk, solve identity and access challenges, and collaborate securely across their supply chain ecosystem.” Other failed platforms faced challenges common to many businesses. A number of platforms emerged before the underlying infrastructure was available to sustain them. For instance, several broadcast streaming or online gaming platforms, including Mpath, broadcast.com, and globalmedia.com, launched around the turn of the century before broadband was widely available. More recently, two digital asset exchanges failed after a short time due to low trading volumes, as the number of people wishing to exchange Bitcoin for other currencies remained too small. Because platforms often brought a new model and structure to existing businesses, they often ran into problems with the legal and regulatory regimes. Uber and Airbnb have famously tangled with local regulators in various cities where they have done business. Other platform companies did not survive regulatory scrutiny. For example, AirPooler and Flytenow, aviation ride-sharing platforms that connected passengers with private pilots who had empty seats on their planes, both shut down in 2015 after the FAA ruled they were engaged in commercial aviation and would have to be regulated as such. In San Francisco, the on-demand valet service Vatler ran afoul of local permitting regulations and had to shut down after barely a year of existence.

As the aggregate data suggests, sources of failure are numerous. Firms fail all the time. Marshall Van Alstyne, Geoffrey Parker, and Sangeet Choudary, for example, offered six reasons that platforms can fail: failure to optimize “openness,” engage developers, share the surplus, launch the right side, put critical mass ahead of money, and have the right imagination. We agree that these are important reasons why platforms do not succeed, but we take a slightly different approach in this chapter that is less general. We focus on four common mistakes that lead platforms to fail as businesses: (1) mispricing on one side of the market, (2) failure to develop trust with users and partners, (3) prematurely dismissing the competition, and (4) entering too late—that is, after a market with very strong network effects and other conditions amenable to a winner-take-all-or-most outcome has tipped. We also explore the most important lessons to learn from companies that came close to winning but never crossed the finish line. What are the lessons of failed platforms that got off the ground, had big starts, but then were unable to tip or hold the market? We also examine the opposite challenge of entering the fray too late. While first movers frequently slip up, that doesn’t mean a manager can sit back and wait. Once a firm overcomes the chicken-or-egg problems and network effects begin to kick in, delay can be deadly.

The bottom line for managers and entrepreneurs is that platforms have to avoid the most common, devastating mistakes, or even those that were early winners can quickly become losers.

Mispricing, on One Side of the Platform

Pricing decisions have been extensively studied by platform researchers, yet managers still get them wrong. The key insight, which we explored in earlier chapters, is that a platform often requires underwriting one side of the market to encourage the other side to participate. Knowing which side gets charged and which side gets subsidized may be the single most important strategic decision for any platform. The challenges become even more acute when rival platforms face intense competition. Good, commonsense pricing may have to be thrown out the window when two or more platforms are racing to create a network effect.

A classic example of a first mover’s failure to appreciate this pricing problem was Sidecar’s missteps in the taxi market. In August 2015, San Francisco–based ride-sharing platform Sidecar announced it would “pivot” to begin focusing primarily on same-day deliveries. It had introduced this service earlier in the year to supplement its main peer-to-peer ride-sharing business. The company anticipated that the majority of revenues going forward would come from deliveries of food, flowers, and even medical marijuana rather than rides, a tacit admission that it had lost its battle to compete with the larger and more well-known ride-hailing platforms Uber and Lyft. That pivot failed to pay off, and the company announced it would suspend operations entirely on December 31. The next month it announced that it had sold the bulk of its assets and licensed its intellectual property to General Motors, which was in the process of developing its own transportation service. Sidecar never became a household name. Its failure was nonetheless significant because Sidecar had pioneered the peer-to-peer ride-sharing model before Uber and Lyft transitioned their start-ups into the space. By 2015, ride-sharing platforms—where smartphone apps connected riders with nonprofessional drivers operating their personal vehicles—had become a pillar of the sharing, or gig, economy, despite the fact that such platforms had only existed for three years.

Sidecar’s cofounder and CEO, Sunil Paul, claimed to have had the idea for such a service as far back as 1998 and received a patent for mobile ride hailing over a wireless network in 2002.5 Paul developed the idea into Sidecar, which he founded in January 2012. The company began beta testing its ride-sharing service in San Francisco the next month and launched the service publicly that June. Riders requested a ride by entering their pickup and drop-off locations through the Sidecar app on their smartphones. A nearby driver would accept the request, and pick up and drop off the rider, who made payment through the app. No cash or other form of payment changed hands. In announcing the launch of the service, Sunil Paul claimed that “Sidecar is more than just the easiest way to get around the city. We have created a platform for the first crowd-sourced transportation network.” Over the next few years, Sidecar expanded slowly and was available in about a dozen cities across the United States by 2015, including Los Angeles, Seattle, Austin, Chicago, Charlotte, Washington, D.C., Brooklyn, and Boston.

On Sidecar’s platform, payment was initially an optional “contribution” to the driver; the app provided information to riders about average donations to help them decide how much to contribute if they chose to do so. Sidecar (and competitor Lyft, which adopted the same payment model) claimed that since they only supplied the technology to connect riders and drivers, they were not a transportation company and should not face the same regulatory requirements as taxis and livery services. Defining payments as donations also allowed drivers to circumvent expanded insurance coverage requirements imposed on commercial vehicles like taxis and limos (although whether insurance companies would see things the same way was unclear). The donation model proved unworkable, and in November 2013, in response to driver feedback, Sidecar dropped it in favor of set fares established when the passenger booked the ride. In announcing the change, Paul said that drivers had “told us loud and clear they would drive more frequently and take longer ride requests if they could reliably depend on fair payment every trip.” Similar to any new platform, Sidecar had to figure out how to price their service to both sides of the market: providers and consumers. Pricing, in this case, meant more than who gets something for nothing and who pays. Ultimately, when platforms face intense competition from other platforms, successful platforms need to attract the critical resources to their platform—and keep them engaged. For livery services, that meant drivers. And as competing platforms emerged beyond the traditional taxi and car services, Sidecar failed to get the pricing right.

Just two months after Sidecar launched, a start-up called Zimride introduced a new peer-to-peer ride-sharing service named Lyft, also based in San Francisco. Zimride had been in business since 2007, operating a long-distance carpooling service that connected riders to drivers using Facebook. Initially marketed to college campuses and businesses as a way to facilitate carpooling for students and employees, Zimride had been largely unsuccessful in marketing its service to the general public. In 2010, Zimride hit upon the idea of local peer-to-peer ride sharing and developed it into Lyft, which launched in 2012. Lyft quickly eclipsed Zimride’s existing long-distance carpooling business and became its main source of revenue and the company’s major focus. Zimride reincorporated as Lyft in March 2013 and then, in July, sold its original Zimride business to car rental giant Enterprise. Meanwhile, things were getting worse for Sidecar: In 2013, Uber introduced its own peer-to-peer ride-sharing service. Uber, also based in San Francisco, had been in business since 2010, when it started as a way to request and pay for a traditional black town car using a smartphone app. By the time Sidecar launched in mid-2012, Uber had expanded to seventeen cities. Its model differed from Sidecar and Lyft in that Uber at that time partnered with existing taxi and car services, so its drivers were professional drivers, including the lower-cost UberX service introduced in July 2012. Recognizing the threat from Sidecar and Lyft, which could offer lower prices because of the reduced licensing and insurance costs faced by nonprofessional drivers, Uber responded. In September, Uber CEO Travis Kalanick told an interviewer, “If somebody’s out there and has a competitive advantage in getting supply, that’s a problem. I’m not going to just let that happen without doing something about it. . . . Uber started out at the high end originally, but the question is can you create a low-cost Uber? Uber has to become a low-cost Uber as well.” In April 2013, Uber announced it would begin offering ride-sharing services from nonprofessional drivers using their personal vehicles in cities where Sidecar and Lyft operated and began rolling out the platform under the UberX name that summer.

Despite its first-mover status, Sidecar expanded much more slowly than its rivals and was eventually squeezed out of the market. In the absence of competition, Sidecar’s strategy might have worked. But the frenetic pace of growth, especially by Uber, eliminated Sidecar’s first-mover advantage. By mid-2015, Uber had expanded to 300 cities around the world and had signed up its one-millionth driver, including over 150,000 active UberX drivers in the United States, and claimed to cover 75 percent of the U.S. population. Lyft had expanded to 65 cities with 100,000 drivers, all in the United States, by March 2015. With Uber and Lyft in the market, competition for drivers and riders was fierce. Both Uber and Lyft aggressively recruited drivers, offering cash bonuses of up to $500 or even $1,000 for drivers who switched from another ride-sharing platform, including Sidecar. Current drivers also received bonuses by referring drivers from another platform. Riders received credits for their first ride and additional credits when they referred other riders. Uber and Lyft periodically cut fares to attract riders. Although the companies claimed that increased ridership would more than make up for the reduced fares in putting money in drivers’ pockets, they took additional steps to avoid alienating drivers when they cut fares. When Uber cut its fares by 20 percent in January 2014, for instance, it reduced its commission on each ride from 20 percent to 5 percent until April, when it raised its commission back to 20 percent on the new, lower fares. Lyft followed suit, dropping its fares by 20 percent in April 2014, and reducing its commission to zero. On and off again subsidies for drivers increased the number available on both Uber and Lyft.

Uber and Lyft both lost money as they pursued aggressive growth strategies and incurred enormous costs just to find and replace drivers. For example, as we noted in Chapter 3, Uber in 2017 lost $4.5 billion despite gross booking revenues of $37 billion. Although Lyft and Uber were primarily targeting each other in their aggressive strategies, Sidecar was caught in the cross fire and attempted to match some of its rivals’ tactics to induce more drivers to use its platform. In early 2015, it was giving $100 bonuses to new drivers, had dropped its commissions to zero in some markets, and offered guaranteed hourly minimums and bonuses for driving during peak Friday and Saturday night hours. But it simply did not have the capital to compete at the scale of its rivals. Once Sidecar fell behind in recruiting drivers and riders, network effects made it extremely difficult to compete.

Lyft and Uber were able to sustain their aggressive growth strategies because they had raised billions of dollars in equity capital. By contrast, Sidecar had raised only $39 million by the end of 2015, creating what Sunil Paul called a “significant capital disadvantage” when announcing that Sidecar was suspending operations in December 2015. He would later write that “the legacy of Sidecar is that we out-innovated Uber but still failed to win the market. We failed—for the most part—because Uber is willing to win at any cost and they have practically limitless capital to do it.” Earlier, when announcing Sidecar’s decision to “pivot” from ride sharing to deliveries in August 2015, he cited the challenge of attracting drivers and riders with relatively little capital: “We were outraised. We have been outgunned in capital from the beginning. Customer acquisition and driver acquisition in this category are very expensive, and it does make a big difference when you offer promotions—$500 for a driver, or $20 credit—versus our more typical $5 offerings of credit.” Sidecar’s failure to keep up with Uber and Lyft in raising capital was a strategic blunder: Management misread the competition, misread the critical role of a supply side (drivers) for a platform market, and missed the logic of network effects. Sidecar deliberately pursued a conservative slow-growth strategy in its efforts to be fiscally responsible. As one early investor commented, “For most companies, the problem they have now is they raised too much money and the valuations are too high. . . . Sidecar was always very disciplined.” Perhaps Sidecar’s fatal flaw was not recognizing the importance of attracting both sides of the platform—drivers and riders. Sidecar focused its energy on innovation—it was the first to introduce several new features such as inputting the destination when requesting a ride, shared rides, and a marketplace model where drivers set their own prices. But this meant that it spent less than its competitors on building a great platform by spending to attract drivers and market aggressively. As one analyst put it, “Sidecar was more of a technology company than it was a marketing company. Its feature set has been very exemplary, but I don’t know that it used the same brainpower to create market share.” Despite being crushed by Uber, Paul offered some grudging respect to his rival: “In some ways, I’m proud of Uber. They were able to take the ride-sharing innovation, rebrand it as UberX, and grow it and scale it like clearly no one else was able to—us or Lyft. It shows that our idea really does work. We’re still proud of the authorship even if it ends up being executed by someone else.” Of course, the jury was still out on Uber, as well as Lyft and other ride-sharing platforms that relied heavily on driver and rider subsidies. Beyond its enormous expenses and financial losses, Uber had to contend with new attempts to impose a minimum wage and regulate the number of drivers. In August 2018, for example, the New York City Council voted to put a hold on the number of licenses granted to ride-sharing firms for one year. In addition, the city imposed rules to assure that drivers were paid a minimum wage. Low compensation for taxi as well as ride-sharing drivers, along with a collapse in taxi medallion values, were causing severe financial hardships for both conventional taxi drivers (including six suicides) and ride-sharing drivers. Uber’s strategy to counter the new regulations was to persuade drivers from Lyft and taxi companies to drive for Uber. It was not yet clear how this move would play out. In the long run, Uber seemed to be following a strategy similar to Amazon’s initial strategy: Get as big as possible as fast as possible, continue to grow or diversify into related businesses that leveraged the same platform, and worry about profits later. In December 2018, Uber also filed documents in preparation for an initial public offering, tentatively scheduled for early 2019 to raise additional cash. If conventional taxi companies and smaller platform competitors went bankrupt or closed down, and if Uber continued to shed its most unprofitable operations, such as in overseas markets, it might well be possible to earn a profit someday. At present, however, Uber lacks a truly profitable division like Amazon Web Services to compensate for its massive subsidies and operating expenses, making profitability harder for Uber to achieve.

Mistrust, Especially in Transaction Platforms

Getting the pricing right is always important. But while getting the price right is necessary in any platform, it is not sufficient for success. Transaction platforms also require two or more parties, who may or may not know each other, to connect. In such a world, building trust is essential. Facebook and LinkedIn, for example, try to build trust by connecting you through friends or business associates; many e-commerce platforms attempt to build trust through rating systems, payment mechanisms, or insurance. In the absence of trust, the players on the platform have to make a leap of faith.

Sidecar was a start-up, and this helps explain why it missed an important platform dynamic. However, eBay had fewer excuses for mistakes that it made, especially in China. eBay was the quintessential e-commerce platform company of the new era: It was the leader in e-commerce platforms at the turn of the century. By 2002, it was the largest online auction site in the world and was the third-ranked website in terms of time spent on the site by users. It generated $1.2 billion in revenue on gross merchandise sales (the total value of sales transacted on the site) of nearly $15 billion in 2002. At the end of the year, the company reported nearly 62 million users and over 638 million items listed for sale on its site. The market value stood at over $21 billion. Yet eBay’s entry into China was another classic failure of a first mover. It is to the credit of Meg Whitman, then CEO of eBay, that she began expanding to international markets in 1999. By 2002, she had set her sights on China, buying a third of the Chinese consumer auction site EachNet for $30 million and then purchasing the rest for $150 million the following year. Whitman rebranded the company as eBay EachNet. Founded in 1999, EachNet more or less imitated eBay in style and content and, at the time of the acquisition, dominated China’s online auction market with more than 80 percent share. (See Figure 4-1.) Given eBay’s success in tipping the market for auctions in the United States, eBay EachNet looked poised to do the same thing in China. It had a dominant share in a rapidly growing space. Moreover, Whitman identified China as a priority for the company’s growth, stating in February 2005 that “market leadership in China will be a defining characteristic of leadership globally.” She later told Time magazine that “China is unique. It is growing rapidly, and it has a tremendous amount of potential, which is why we have made it a priority for the company.” By the end of 2006, eBay had invested $300 million in its Chinese operations, yet it had essentially admitted defeat and yielded the Chinese online market to its major rival.

Why did eBay, the first mover, fail? As in the case of Sidecar, the story begins with competition and getting the pricing right. When eBay entered China, Alibaba was the largest player in the Chinese e-commerce world. Founded in 1999, Alibaba was, like eBay, an online marketplace. It did not own and sell inventory; rather, it facilitated transactions between third-party buyers and sellers, taking a commission on each sale. Alibaba differed from eBay in that it focused on business-to-business (B2B) transactions, providing a platform for small and medium-sized enterprises in China. eBay, on the other hand, focused primarily on the consumer-to-consumer (C2C) market. Despite the distinction, Alibaba’s CEO Jack Ma saw eBay’s entrance into China as a major threat to Alibaba’s business. He recognized that the distinction between B2B and C2C commerce on the Internet was a blurry one. If eBay gained a foothold in the C2C market, it would be a short step for small businesses to list their wares on eBay to sell to both consumers and other small businesses and eat into Alibaba’s business.

To respond to the challenge posed by eBay, Ma and Alibaba took on the American firm head-on, launching a rival C2C marketplace called Taobao (“hunting for treasure”). The two auction sites were initially identical in many ways. In fact, Taobao’s first web pages were copied directly from eBay’s design. But Alibaba quickly pivoted, recognizing that one cannot win a platform battle with a me-too product. Perhaps Jack Ma’s first critical decision was to make Taobao’s service free to buyers and sellers; unlike eBay, it did not charge listing fees or commissions. Ma was willing to lose money on Taobao, funding the site through cash generated by Alibaba’s other businesses and concentrating on building market share. eBay, at least initially, followed its existing model of charging listing fees and commissions on each transaction. In addition, Taobao did not emphasize the auction model but rather direct sales: Only 10 percent of its listings were auctions, compared to 40 percent for eBay EachNet. Alibaba/Taobao recognized, in the words of one executive, that “what our customers really wanted, we realized, was simply a storefront for selling their products.” There was another difference: While eBay initially did not let buyers and sellers interact directly (perhaps fearing they would take their transaction off-line and avoid paying eBay’s commission), Taobao encouraged buyers and sellers to communicate with each other and added instant messaging to the website, allowing buyers and sellers to build trust.

For any transaction platform to be successful, trust is essential. Online payments represented another important area of difference. In 2004, Alibaba launched its own online payment system called Alipay, similar to PayPal, which eBay had acquired shortly before entering the Chinese market. Alipay, unlike PayPal, used an escrow model. At the time of a transaction, funds would go into an escrow account and be released only after the buyer had received and inspected the item, which was important in a culture where trust and security were an issue. In addition, at a time when only a small fraction of Chinese consumers had credit cards, Alipay also formed partnerships with leading banks and the China Post so that customers without a debit or credit card could fund their Alipay accounts with cash at one of its 66,000 offices. Alipay also exploited the fact that a Chinese-owned payment processing system faced far fewer regulatory restrictions from the Chinese government. Despite eBay’s advantages, including its big balance sheet, global e-commerce platform, and the fact that EachNet was the leading online C2C auction site in China, Taobao had surpassed eBay on a variety of measures by 2005. By August 2005, Taobao was reaching 15,800 out of every 1 million Internet users in China, compared to under 10,000 for eBay. User satisfaction was also higher for Taobao than for eBay—77 percent to 62 percent. In the first quarter of 2005, Taobao’s gross merchandise volume (GMV) surpassed eBay’s for the first time—$120 million compared to $90 million for eBay, a remarkable feat considering that eBay’s GMV was ten times that of Taobao at the beginning of 2004.31 eBay suffered a further blow in August 2005 when Yahoo invested $1 billion in Alibaba and handed over its China operations to Alibaba in return for a 40 percent stake in the company. The capital infusion obviously strengthened Alibaba in its fight with eBay, in part by enabling it to continue to offer Taobao’s platform for free. At the time, eBay pledged to continue the fight, with Whitman calling China “a perfect market for eBay.” By March 2006, however, Taobao had become the market leader in China’s online C2C market with a 67 percent market share, and Alibaba CEO Jack Ma declared that “the competition is over.” eBay admitted defeat a few months later when, in December 2006, it effectively withdrew from China, shutting down its Chinese website while forming a joint venture with Hong Kong–based Internet portal TOM Online, which had less than 10 percent share of the Chinese market by 2010. Why did eBay fail in China? In retrospect, it appears that eBay did many things wrong compared to Taobao. For example, eBay integrated EachNet into its global platform and global brand identity, reproducing the look and feel of eBay’s website and eliminating localized features popular with Chinese consumers. Chinese consumers appeared to prefer Taobao’s busier look and feel and found eBay’s site too plain and minimalistic. eBay also failed to give enough power to local executives, who complained that executives in California did not listen to them and directed operations from San Jose rather than trusting local leaders. eBay also moved aggressively to sign exclusive partnership deals with China’s largest Internet portals, effectively shutting out Taobao. But Jack Ma and Taobao relied on TV, print ads, billboards, SMS websites, and word of mouth, which proved to be more effective and less expensive strategies for bringing users to the site. As Ma said in 2004, “The cost for eBay and international companies to come to China is so high. They spend $100 million dollars, we spend $10 million, but the effect is the same.” Although many of these failures are symptomatic of any global expansion, the biggest lessons from eBay are common to many failed platforms. As the CEO of eBay China admitted to us in an interview, “eBay’s single biggest problem in China was trust.” Developing a new platform requires buyers and sellers to trust each other; and successful platforms build mechanisms to solve this problem. eBay relied on PayPal, which initially acted as it did in the United States: It was designed as a payment system, very much like a bank. For Chinese consumers, unfamiliar with e-commerce, that was not enough. Since Alipay used an escrow model (which did not release payment until the consumer was satisfied) and the existing financial system, Taobao neutralized eBay’s early mover advantage.

Finally, eBay’s failure to price “right,” to offer its platform for free, was also crucial. As Jay Lee, eBay’s managing director for Asia-Pacific, put it several years later, “It’s very hard to compete with free.” In early 2006, eBay did begin offering free listings and integrated escrow services into PayPal, but by then it was too late. Alibaba had quickly become the default destination for consumers, which attracted more suppliers, which attracted more consumers. In other words, the network effects kicked in. In the end, eBay’s experience in China seemed to validate one of Alibaba CEO Jack Ma’s favorite aphorisms. Speaking of Western Internet powerhouses like eBay and Yahoo, he said that “they are the sharks in the ocean, we are the crocodiles in the Yangtze River. When they fight in the Yangtze River, they will be in trouble. The smell of the water is different.” Hubris, or Dismissing the Competition

One common misconception among players in a platform market is that, once the market tips in your favor, you will be the long-run winner. Often this is true. But there is a better way to think about markets that have tipped: It is the winner’s opportunity to lose. Hubris, along with overconfidence and arrogance, to name a few misdirected traits, can produce spectacular failures—even in markets that look like winner-take-all.

In 1998, we wrote a book called Competing on Internet Time: Lessons from Netscape and Its Battle with Microsoft. We explored how Netscape, a first mover with an early and overwhelming market-share lead, managed to lose the browser wars to Microsoft. Browsers were a classic innovation platform: Webmasters had to optimize their websites to exploit key features in a browser, while application developers took advantage of a browser’s APIs. When Netscape dropped from 90 percent market share in 1996 to virtually zero, and Microsoft’s Internet Explorer captured close to 95 percent by 2004, most pundits proclaimed the browser wars were over, the market had tipped, and Microsoft had won. It would require a monumental screwup for Microsoft to lose the market, but this is exactly what happened.

Perhaps the good news is that it took Microsoft almost a decade to lose its leading position. The failure came in two chapters: extremely poor product execution between 2004 and 2008; and then inferior product innovation between 2008 and 2015.

Internet Explorer and Firefox, 2004–2008: With its 95 percent share in 2004, Microsoft’s competition primarily came from Firefox—the open-source Mozilla project. This browser originated when Netscape made the source code for Navigator available in early 1998 to any interested developer who wanted to download and modify it. While Netscape’s decision to go open source was in some ways an admission of defeat, it also represented an effort to ensure a viable alternative to Internet Explorer. Network World put it this way when the project launched: “Netscape’s source code giveaway will create the market conditions necessary for sustaining a competing browser in a Microsoft-dominated world.” Netscape reported more than 200,000 downloads of the source code within the first two weeks of making it available. Developers focused mostly on adapting the software for different operating systems, fixing bugs, and improving security, which could happen relatively quickly with a large number of developers working on the code. Mozilla introduced its first browser in June 2002, with innovative features such as tabbed browsing, the ability to select a word or phrase and search the web for the selected text, and the ability to store common information for automatically filling in forms. The next round of the browser wars truly opened with Mozilla’s release of Firefox in September 2004. Seeking to avoid software bloat and performance issues, two developers working for the Mozilla nonprofit foundation began what would become the Firefox browser in late 2002. Their goal was to produce a stand-alone browser that would be faster, simpler, and more secure. After its release, Firefox quickly began to eat into Internet Explorer’s market share. One source reported that Firefox had 4.5 percent of the browser market by the end of November 2004, while Internet Explorer’s share had dropped 5 percentage points to 89 percent. By the end of 2007, Firefox had grabbed 17 percent of browser usage, compared to 76 percent for IE. Why was Microsoft slipping so fast?

In late 2004, Computerworld wrote that “with no serious competitor, Microsoft stopped development of IE. New versions appear from time to time, but it has been years since IE offered groundbreaking new features. Meanwhile, development of other browsers has continued to the point where many consider them preferable to IE in performance, security, ease of use, added features and even help desk support.” It went on to declare that all of its competitors—Firefox, Navigator, and Opera—had more features, were more secure, and were available on more different operating systems. In terms of speed loading pages, the article noted that all the other browsers “beat IE by a country mile.” In effect, Microsoft took its dominant position for granted. Internet Explorer 6 was released in 2001 and Microsoft did not introduce a new version, other than for bug fixes and security patches, until 2006, opening the door to competitors. Security was a particular problem: Hackers and security experts continually found flaws, after which Microsoft would release a patch to fix the vulnerability, only to see new vulnerabilities uncovered. One issue was that Microsoft had integrated IE into the operating system, enabling the browser to execute Windows code. This integration made it possible to create web-based applications with richer features. The downside was that doing so created abundant opportunities for hackers to execute viruses and other forms of malware on users’ machines, resulting in “a security nightmare for IT organizations.” Security became such a problem that the U.S. Computer Emergency Readiness Team, noting that important components of IE were “integrated into Windows to such an extent that vulnerabilities in IE frequently provide an attacker significant access to the operating system,” recommended using a different web browser as a way to reduce vulnerabilities. PCWorld declared in 2006 that IE 6 “might be the least secure software on the planet,” and named it number eight on its list of the twenty-five worst tech products of all time. Microsoft admitted that it had failed to adequately shepherd its browser. The head of the Internet Explorer team noted in March 2006: “I want to be clear: We messed up. We messed up. As committed as we are to the browser, we just didn’t do a good job demonstrating it.” Bill Gates also acknowledged that Microsoft had left too long of a lag time between browser updates. Microsoft responded by accelerating the timetable for IE 7. Despite its release in 2006, Firefox continued to gain ground, accounting for about 32 percent of web traffic by the end of 2009, while IE’s share fell to 56 percent. Firefox’s share remained flat for the next year or so, and then began a slow decline that left it with only 5 percent of the market by early 2018. Its losses were not IE’s gain, however, as the browser wars entered a new phase with the emergence of a new competitor in late 2008, Google’s Chrome.

Google Chrome, 2008–2016: Google launched Chrome in September 2008. It gradually gained traction in the market and soon began gaining ground on IE and Firefox, turning the browser contest into a three horse race. By one measure, total web traffic using Chrome passed Firefox at the end of 2011 and overtook IE in the middle of 2012. Other measures of users show Chrome passing Firefox in mid-2014 and reaching a near tie with IE by early 2016. (See Figure 4-2.)

The trend was clear: Chrome rapidly took market share from both Firefox and IE in the years after its launch. When Google released the first version, it boasted that it had developed a new JavaScript engine that could execute JavaScript code ten times faster than competing browsers. Wired reported in September 2008 that the first version of Chrome ran some benchmark tests ten times faster than Firefox and Safari and fifty-six times faster than the current version of Internet Explorer, IE 7. That’s right: fifty-six times! Chrome was also developed with a multi-process architecture, when meant that each browser tab or plug-in ran as a separate process, so that if a web page or application crashed, it would not bring down the entire browser. In addition, Chrome was sandboxed, meaning that code executed in the browser could not access the underlying OS, reducing security threats of the kind that had so plagued Internet Explorer. Chrome was also the first to integrate the search function into the main address bar and had a much more minimalistic and streamlined look and feel. Chrome’s rapid development cycle—it released its seventeenth major version in early 2012, at which time IE was on version 9 and Firefox stood at version 10, despite their much longer life spans—meant that new features and fixes were available sooner. Chrome’s automatic update process, which Wired called “reliable, consistent, and effective,” worked to quickly move the vast majority of users to the current version. PCWorld, for instance, declared Chrome the best browser in February 2012, citing its speed in loading pages and executing JavaScript and HTML5 code, security, and availability of add-ons and extensions that expanded its functionality. What accounts for Chrome’s rapid ascent, and IE’s rapid demise? How can a market that supposedly tipped become unraveled? When Microsoft won the first browser war, it delivered the best product and then used its market power with Windows to create the best browser platform and drive competitors out of the market. By assuming it had won, by dismissing the threats from competitors, and by failing to innovate, Microsoft created an opening for Mozilla and then Google to exploit. While IE remained slow and buggy, with a reputation for poor security, Chrome offered speed, security, stability, and rapid development cycles (which enabled it to get fixes and new features to market quickly). There is a critical lesson to be learned here: Although network effects are very powerful, they do not guarantee success forever. It took a long time for Microsoft to fail, but fail it did.

Mistiming, or Failure to Act Before the Market Tips

Perhaps the most classic platform mistake is mistiming the market. In the browser wars, late movers were able to build strong platforms, but mostly because of the incumbent’s letdowns. If Netscape had done some things differently, it might have survived the onslaught from Microsoft’s Internet Explorer. Then, if Microsoft had kept Internet Explorer competitive, Firefox and Chrome probably would have failed to build credible competing platforms.

The smartphone market is a classic example of how great products plus all the resources in the world may still lead to failure in a platform market when entry is too late. Here again, Microsoft is the poster child for failure. In 2017, despite billions and billions of dollars of investments over a decade, Microsoft’s Windows Phone was dead. It missed the platform window and never recovered.

When Apple released the iPhone in July 2007, it faced several entrenched players in the smartphone OS platform market, including Microsoft, Research in Motion’s BlackBerry, and Nokia’s Symbian. The last was the runaway market leader, accounting for over 63 percent of smartphones sold in 2007. Microsoft had 12 percent of the market that year, while BlackBerry had a 10 percent share. In ways that were not fully clear at the time, the iPhone revolutionized the smartphone category, leaving the entrenched players struggling to catch up. None did. By 2014, Symbian had disappeared and BlackBerry had all but vanished, its share measured in tenths of percentage points. Apple’s iOS captured most of the handset industry’s profits, but only about 15 percent of the unit volume in 2015.

The clear market-share winner in the global smartphone platform race was Android, the open-source mobile platform Google introduced in 2008. When Google announced Android in late 2007, existing smartphone platform leaders did not realize the implications. A spokesman for Nokia said, “We don’t see this as a threat.” Symbian’s head of strategy told Reuters, “We have been going nine years and have probably seen a dozen new platforms come in and tell us we are under attack. We take it seriously, but we are the ones with real phones, real phone platforms and a wealth of volume built up over years.” An executive for Microsoft’s Windows Mobile project similarly dismissed the challenge: “It really sounds that they are getting a whole bunch of people together to build a phone and that’s something we’ve been doing for five years. I don’t understand the impact that they are going to have.” The impact turned out to be significant. With over 80 percent of the worldwide market in 2018, Android was clearly the dominant smartphone platform. The smartphone OS game became a two horse race, with other platforms—including Microsoft’s—consigned to irrelevance.

The combined impact of the iPhone and the emergence of Android was to usher in a radically new phase in the evolution of mobile phones. Microsoft was caught flat-footed. Steve Ballmer famously gave an interview a few months after the release of the iPhone, saying, “There’s no chance that the iPhone is going to get any significant market share. No chance.” Not perceiving a serious threat, Microsoft took three full years to release its rebooted Windows Phone 7, its first attempt to move into the post-iPhone world of touch-screen‒centric user interfaces for smartphones. By the time the new Windows Phone was released in 2010, iOS and Android had nearly 40 percent of the global smartphone market between them and would reach a combined share of over 65 percent in 2011. Despite very positive reviews of Windows Phone 7 as well as Nokia’s decision to abandon Symbian and join forces with Microsoft to create a third smartphone ecosystem, it was too late. The smartphone platform market had tipped into a stable duopoly, which made it difficult for a third viable platform to emerge.

Early reviews of Windows Phone 7 were very positive. ArsTechnica declared that Windows’ new look made “iOS look dated, and Android cluttered and fussy.”57 PC Magazine declared Windows Phone 7 “a very impressive effort,” with “a great-looking new design” and “powerful Office productivity and Zune [Microsoft’s digital media platform] entertainment features.” It predicted that, while it might take a few sales from the iPhone, “it was more of a threat to the chaotic Android consumer experience.” Engadget in its review concluded that, with “a fast and smart method of getting around the OS, great Office and email experiences, and a genuinely beautiful and useful user interface, Microsoft has definitely laid the foundation for the next several years of its mobile play.” Reviews for subsequent versions of the Windows Phone OS continued to be positive, praising its smooth and fast performance, user interface, and Internet Explorer’s web browsing capabilities. Wired gave Windows Phone 8, released in October 2012, a score of 8 out of 10, declaring that “the third mobile platform is finally, really here.” The reviewer for the Verge wrote that “it’s not a stretch to say that Windows Phone 8 has the best home screen—the perfect combination of flexibility, design, and simplicity—of any major platform right now.” The positive reception afforded Windows Phone after its launch in 2010 gave observers cause for optimism. One reviewer pointed out that most mobile phone users were not yet smartphone users in 2010, so there was a large untapped market, not yet committed to iOS or Android, that Microsoft had a chance to win over with Windows. As one reviewer put it, “Yes, Microsoft is late to the game, but it’s a game that’s still in its early stages.” The platform got another potential boost in 2011 when Nokia, at the time the world’s largest handset maker, decided to abandon its own Symbian OS and adopt Windows Phone for its devices. Nokia CEO Stephen Elop predicted it would become a third—after iOS and Android—viable integrated ecosystem: “Today, developers, operators and consumers want compelling mobile products, which include not only the device, but the software, services, applications and customer support that make a great experience. Nokia and Microsoft will combine our strengths to deliver an ecosystem with unrivaled global reach and scale. It’s now a three horse race.” Some suggested the partnership might revitalize Windows Phone’s flagging fortunes: “Windows Phone 7 is no one’s priority. But now Microsoft has a leading vendor committed to use the platform.” At a time when Windows Phone had about 2 percent of the smartphone OS market, market research firm IDC predicted that Windows Phone would reach 20 percent market share in 2015, surpassing iOS for second place behind Android. And another firm, Pyramid Research, believed that Windows would be the market-share winner! Such predictions failed to materialize. In a desperate effort to keep sales of Windows Phone alive, Microsoft even acquired Nokia’s devices business in 2013 for $7.2 billion. It was forced to write off the entire value of the acquisition two years later. Although its partnership and acquisition of Nokia represented an attempt to integrate hardware and software along the model Apple pursued with the iPhone, Microsoft continued to license Windows Phone to other vendors. Not surprisingly, major players like Samsung, HTC, and LG put little effort into the platform. By the end of 2015, Windows Phone sales had fallen off a cliff, to under 3 percent of the market.

Why did an excellent product, backed by the largest cell phone manufacturer in the world, fail so miserably? The answer is that platform competition is fundamentally different from product competition. As we have said before, in a platform market, it is the best platform, and not the best product, that usually wins. Google Android had already lined up the majority of handset makers, application developers, and users. Apple also had a fiercely loyal following of users and application developers. The intense drive of the incumbents (Apple and Google), bolstered by the power of network effects driven by the number of users and high-quality complements, and the difficulty of catching up with users and application developers after a late start, probably doomed Microsoft’s efforts from day one.

One persisting challenge was the relative dearth and lower quality of apps for Windows Phone compared to iOS and Android. As PC Magazine noted in 2010, “The battle is now shifting to apps. Windows Phone 7 has the basic features to make it in the marketplace. But Apple has a massive head start, and Android has huge market share which could make it appealing. If Windows Phone can draw third-party developers over from Android, it could sap some of that platform’s energy.” Quality was also an issue. Engadget, despite its otherwise glowing review of Windows Phone 7, reported that for many apps “the results seemed second rate in comparison to the same applications on other platforms” and concluded that “the OS clearly needs time to mature, and developers will have to work a lot harder to get their apps up to spec with the competition.” Windows, it concluded, “is a good year behind market leaders right now, and though it’s clear the folks in Redmond are doing everything they can to get this platform up to snuff, it’s also clear that they’re not there yet.” Two years later, Windows was still struggling to catch up.

The simple fact was that developers could reach well over 90 percent of the smartphone market by building apps for iOS and Android. It was rarely worth the time and effort to develop for a platform with less than 3 percent market share. As one analyst put it in 2012, “Developers go where the money is, and the money is where people are.” In an effort to overcome that challenge, Microsoft at times paid app developers to write apps for Windows Phone, at a cost estimated at between $60,000 and $600,000 for each app in 2012. For example, Microsoft underwrote the cost of developing a Windows Phone version of the social networking app Foursquare in 2012, which the company would probably not have done otherwise. As Foursquare’s head of business development put it, “We have very limited resources, and we have to put them toward the platforms with the biggest bang for our buck.” In absolute terms, the number of Windows Phone apps available grew rapidly after its launch in 2010. From a base of only 1,000 apps available when Windows Phone 7 launched, the platform grew to 8,000 at the time of the Nokia agreement in 2011, to over 100,000 when Windows Phone 8 launched in 2012, and 320,000 by the end of 2014. Even with this growth, and despite creating a potentially potent hybrid platform, the Windows Phone ecosystem remained far smaller than competitors. Android and iOS respectively had 1.4 million and 1.2 million apps available by late 2014. While the sheer volume of apps was itself a major issue, the availability of popular apps was even more important. Instagram, for example, was not available on the Windows Phone despite the fact that, at the time, it had been available for iPhones for two years and on Android for six months. One reviewer concluded that “it’s not like the store is empty—there are more than 120,000 apps available. They just aren’t the apps you want.” Many popular apps lagged far behind the launch date on iOS and Android. The music streaming service Pandora became available for Windows more than four years after it first launched on iOS. The ride-hailing app Uber came to Windows Phone over four years after its initial launch. The Windows Phone app for the social networking site Tumblr arrived nearly three years after it was first released, while the music streaming service Spotify was a little over two years behind the other platforms. The popular mobile messaging app Snapchat, which first launched in mid-2011, did not make an authorized version available for Windows Phone until early 2016. And the list went on.

By 2017, Windows Phone was irrelevant. As one analyst described the situation: “Windows Phone is dead. Lumia can fade away. But Microsoft has already moved on” to new areas, such as the cloud. Key Takeaways for Managers and Entrepreneurs

In this chapter, we discussed four of the most common mistakes that platform ventures make. These failures were driven by mispricing and inadequate subsidies to get network effects going, not building sufficient trust with platform users, not paying sufficient attention to competitors, and entering markets that have already tipped, without a viable strategy or understanding of why they tipped and what might overturn the status quo. So, when it comes to avoiding mistakes in platform businesses, what are the key takeaways for managers and entrepreneurs?

First, since many things can go wrong in a platform market, managers and entrepreneurs need to make concerted efforts to learn from failures. It is easy to say, “I would never do something stupid like that.” Yet, with platform strategy, the challenges are more complicated to identify because there are so many moving parts. Firms not only have to coordinate internal operations, a supply chain, and novel methods of distribution. They also have to manage complements, overcome chicken-or-egg problems, and simultaneously stimulate multiple sides of a market. As we observed up front, platforms have high failure rates, like most start-up ventures. Despite the huge upside opportunities that platforms offer, pursuing a platform strategy does not necessarily improve the odds of success as a business.

Second, since platforms are ultimately driven by network effects, getting the prices right and identifying which sides to subsidize remain the biggest challenges. Uber’s insight (and Sidecar’s failure) was recognizing the power of network effects to drive volume by dramatically lowering prices and costs on both sides of the market. However, as we have noted, this strategy requires lots of venture capital or other sources of cash to sustain. In fact, most though not all of the most successful platforms we studied built their businesses on low-cost capital, which allowed the winners to aggressively acquire at least one side of a market. In addition to Uber, think YouTube (which was free for producers and content consumers for several years), or Facebook (which offered free access to its social network). Of course, outside funds are never free forever: Market-share winners still have to pay back investors eventually. Uber has struggled mightily to stem its losses, such as by selling money-losing operations overseas. It might have a successful IPO in 2019 because of its continuing rapid growth, though it may still be years away from making a profit, if it ever does. Nonetheless, other platforms, such as Google, Facebook, eBay, Amazon, Alibaba, and Tencent, successfully made the transition from losing money to making money as they gained scale and expanded until they found a business or a business model that made profits. Clearly, this is what investors expect of Uber.

Third, it is important to put trust front and center. Virtually all platforms require trust because they are intermediaries linking users and other market participants that generally have no personal interactions or relationships. Asking customers or suppliers to take a leap of faith, without history and without prior connections to the other side of a market, is usually asking too much of any platform business. eBay’s failure to establish mechanisms for building trust in China, like Alibaba did with Taobao, is an error that platform managers can and should avoid.

Fourth, and this may sound obvious, but timing is crucial. Being early is preferable, but no guarantee of success; being late can be deadly, unless your competitors screw up. We have all sat in meetings with managers where they tell their boards and staff that they have the “best” product on the market, which is surely going to win. The problem is often timing: The best platform usually trumps the best product, but platforms and their ecosystems take time to build. This is especially true for innovation platforms, where you need armies of developers to innovate, and you need many consumers to attract those developers and then use those innovations. Come too late to that party, as Microsoft did with Windows Phone, and you are bound to fail, even with great products and virtually unlimited financial and engineering resources. It is essential that challengers carefully probe why a market has tipped and what strategies and investments might disrupt the platform leaders.

Finally, hubris can lead to disaster. Dismissing the competition, even when you have a formidable lead, is inexcusable. One of the greatest dangers in business is complacency. Andy Grove, former CEO of Intel, had a wonderful saying: “Success breeds complacency. Complacency breeds failure. Only the paranoid survive.” The dangers of success breeding complacency are perhaps no greater than in platform markets because of network effects. Once a market seems to have tipped, it is easy for managers to relax and believe that the tipping is permanent. Yet, in fact, it is extraordinarily hard in many tipped markets for competitors to build a viable platform business. Complacency, even among successful platform companies, can indeed breed failure—which Microsoft proved in the browser market during the 2000s, after it beat Netscape in the 1990s. If you cannot stay competitive, no market position is safe. A senior executive from Google even told us that he’d like to “erect a statute to Steve Ballmer” for performing so poorly in browsers and smartphones.

Traditional non-platform firms have an especially difficult challenge in the digital age. But even they should not simply give in to new platform competitors. How old dogs can learn new tricks—a challenge that is difficult but possible—is the subject of the next chapter.