Digital platforms: High valuations, but high risk of failure

The world’s largest companies are digital platforms. Amazon, Apple, and Microsoft have each temporarily surpassed $1 trillion in value. Uber’s upcoming initial public offering is expected to value the 10-year-old firm at $90 billion – roughly the same value as Dow DuPont, a conglomerate of two firms founded in the 19^th century.

Companies from startups to enterprises have tried to emulate the rapid success of platforms, which achieve faster growth with fewer employees and more R&D spending than the non-platform firms. But these companies have often learned the hard way that, despite being easy to build and easy to scale, platforms are difficult to sustain.

“There are hundreds – and across the world thousands – of platforms being created without the thought to the economics or the strategies that make them work,” said Michael A. Cusumano, MIT Sloan professor and co-author of “The Business of Platforms: Strategy in the Age of Digital Competition, Innovation, and Power,” out May 7. “The strategy is more complicated. The business model is more complicated.”

2 types of digital platforms

Digital platforms have existed for more than 30 years. One of the earliest examples is the Windows operating system. On its own, it has little to no value, Cusumano said. The real value comes from the many third-party products and services that run on Windows.

That makes Windows a type of innovation platform, according to Cusumano and co-authors Annabelle Gawer and David B. Yoffie. Other innovation platforms include the Google Android and Apple iOS mobile operating systems, along with the Amazon Web Services and Microsoft Azure cloud computing platforms.

The authors also define a transaction platform as an entity that enables the exchange of goods and information. These services, which include Amazon Marketplace, Google Search, Uber, and social media platforms, reduce the friction between those with something to sell and those who are looking to buy.

There are also hybrid platform companies, which rely on both innovation and transaction platforms within the same firm. Companies typically begin with one type of platform before they extend into the other category. Facebook, for example, began as a social networking site (a transaction platform) before introducing the capability for third parties to develop plug-ins such as games (an innovation platform).

When platforms succeed, they stand out. “The Business of Platforms” analyzes the performance of 43 digital platform companies from 1995 to 2015 that arose with the personal computer, the internet, and mobile devices such as smartphones. Compared to non-platform firms in the Fortune Global 2000, the platforms have higher sales growth, market value growth, and operating profits. They also achieved this growth with almost half as many employees – fewer than 10,000 on average, compared to 19,000 for the non-platform firms – and greater spending on research and development.

“We pay disproportionate attention to these platforms because they are the most valuable companies in the world,” Cusumano said. “That reality has drawn so many entrepreneurs and investors – but it takes a lot of work to get to that point, and they may not realize how hard it is.”

6 ways that platforms can fail

The vast majority of digital platforms fail. And they fail quickly – in less than five years, according to the authors’ analysis. Business-to-business marketplaces, Web browsers, and ride-sharing platforms fare particularly poorly, typically lasting less than three years.

“Failure is not random. It’s not the same as conventional markets,” Cusumano said. “It’s largely about figuring out which side of the market is more important.”

Six types of poor decisions can cause a digital platform to fail:

Getting the pricing wrong. The three major video game consoles – Nintendo Wii, Sony PlayStation, and Microsoft Xbox – opted to sell their hardware at near cost and charge more to game developers to have their games on the console, in an effort to boost revenue from game sales. However, the low cost of consoles meant that consumers simply bought more than one system, which also meant that developers in turn could afford to produce games for more than one system. As a result, none of the consoles have achieved market dominance.

Failing to establish trust. eBay entered China with a flourish in the early 2000s, only to withdraw from the country by 2007. Alibaba’s consumer-to-consumer marketplace, Taobao, used an automated payment system built on an escrow model; it also used a brand that Chinese consumers recognized.

Dismissing the competition. Microsoft’s Internet Explorer plummeted from 95% market share in 2004 to 5% in 2016. The company fell flat twice, first by failing to provide feature and security updates (losing share to Firefox) and then by failing to update regularly (losing share to Chrome).

Getting the timing wrong. After Apple released the iPhone in 2007, it took Microsoft three years to update the Windows Phone. By that time, iOS and Android were the clear mobile operating system leaders.

Failing to solve the chicken-or-egg problem. General Electric has spent several years and several billion dollars developing Predix, its operating system for the Internet of Things, but has struggled to get companies outside its own customers to adopt.

Entering a low-margin business. Ride share companies lose money because they subsidize riders with low fares and discounts, and because they have to pay drivers. “It seems like a magical digital business – two sides of the market that are under-supplied – but introducing a platform doesn’t make a bad business magically profitable,” Cusumano said. That’s why Uber, Lyft, and China’s Didi Chuxing have made significant investments in autonomous vehicles (as well as bicycles and scooters) to eliminate their enormous driver payments.

Other key lessons from big digital platform winners

Platform success is by no means guaranteed, but there are some key lessons for companies looking to get into the platform business, Cusumano said.

AirBnB, for example, links people with a need (a short-term rental) to an asset that fills that need (an underutilized home) with little fixed costs. “It’s almost like free money,” he said.

Meanwhile, operating systems benefit from economies of scale as well as powerful network effects whereby platform participants increasingly benefit as there are more participants in the platform ecosystem.  For example, whereas Uber and Lyft have to replicate the service they provide with every single ride, Microsoft, Apple, and Google only need to build a version of Windows, iOS, or Android once. “The cost to them is the same whether they sell it once or a million times,” Cusumano said.

Moreover, the more applications there are for the operating systems, the more users are drawn to the OS platform – an indirect network effect.  Microsoft, Apple, and Google do not have to pay developers to write these millions of applications, which make the platforms extraordinarily useful. In the case of Uber, there are also network effects as more and more drivers attract riders and vice versa. But Uber has to pay a lot of money to get and keep drivers, and it has to keep prices below a taxi to get riders. So there are network effects with Uber, but the company also loses big and has little chance of making a profit in the forseeable future, Cusumano said.

Prudent platforms also need to stay a step ahead of potential regulatory challenges. As Amazon grew, it entered agreements to collect state sales tax before regulators could force the company’s hand. On the other hand, Facebook, Twitter, and YouTube faced scrutiny from American and European regulators for failing to remove hateful or inappropriate content.  

While digital platforms have clear differences from traditional businesses, a successful platform is often built on the same foundation as a successful business, Cusumano said.

“Platform businesses are everywhere around us, and they affect our personal and professional lives almost completely – but platformizing a bad business does not make it a good business,” he said.