Opinion
Computing Applications Economic and business dimensions

Three Paradoxes of Building Platforms

Insights into creating China's Taobao online marketplace ecosystem.
Posted
  1. Introduction
  2. The Control Paradox
  3. The Weak Partner Paradox
  4. The Killer App Paradox
  5. Author
Three Paradoxes of Building Platforms, illustration

Platforms have apparently become the holy grail of business on the Internet. The appeal is plain to see: platforms tend to have high growth rates, high entry barriers, and good margins. Once they reach critical mass, they are self-sustaining. Due to strong network effects, they are difficult to topple. Google and Apple are clear winners in this category.

Taobao.com—China’s largest online retailer—has become such a successful platform as well. Taobao merchants require myriad types of services to run their storefronts: apparel models, product photographers, website designers, customer service representatives, affiliate marketers, wholesalers, repair services, to name a few. Each of these vertical services requires differentiated talent and skills, and as the operational needs of sellers change, Taobao too has evolved, giving birth to related platforms in associated industries like logistics and finance. Although it has become habit to throw the word "ecosystem" around without much consideration for its implications, Taobao is in fact a thriving ecosystem that creates an enormous amount of value.

I have three important lessons to share with future platform stewards that can be summarized in three fundamental paradoxes. These paradoxes encapsulate the fundamental difficulties you will run into on the road toward your ideal platform.

Back to Top

The Control Paradox

Looking back at Taobao’s success and failures over the past 10 years, I have come to believe that successful platforms can only be built with a conviction that the platform you wish to build will thrive with the partners who will build it with you. And that conviction requires giving up a certain amount of control over your platform’s evolution.

People are used to being "in control." It is a natural habit for most people to do things on their own, and by extension command-and-control has become a dominant way of thinking for most businesses. But platforms and ecosystems require completely new mind-sets. Being a platform means you must rely on others to get things done, and it is usually the case that you do not have any control over the "others" in question. Yet your fate depends on them. So especially in the early days, you almost have to have a blind faith in the ecosystem and play along with its growth.

Taobao began with a belief in third-party sellers, not a business model where we do the buying and selling. On one hand, it was our task to build the marketplace. On the other hand, Taobao grew so fast that many new services were provided on our platforms before we realized it. In a sense, our early inability to provide all services on our own caused us to "wake up" to an ecosystem mind-set, and the growing conviction that we had to allow our platform to evolve on its own accord.

Despite such strong beliefs, however, there were still many occasions when our people wanted to be in control, to do things on their own, and in the process ended up competing against our partners. When such things happen, partners start to doubt whether they can make money on your platform, and this may hamper ecosystem momentum. For example, we once introduced standard software for store design, hoping to provide a helpful service and make some extra money. However, it soon became apparent that our solution could not meet the diverse needs of millions of power sellers, and at the same time, it also impacted the business of service providers who made their living through sales of store design services. Later, we decided to offer a very simple basic module for free and left the added-value market to our partners.

What makes this paradox particularly pernicious is the fact that working with partners can be very difficult, especially when the business involved grows more and more complex. In the early days of a platform, customers are often not happy with the services provided. Platform leaders have to work very hard to keep all parties aligned and working toward the same goal. If the platform decides to take on responsibilities itself, it will stifle growth of the ecosystem. Yet the incubation period is a long and difficult process, and requires a lot of investment. Without strong conviction, it is very difficult to muddle through this stage, which may take years.

In very specific terms, you must sell your vision of a third-party driven service ecosystem to the end user when you have very few partners (or even no partners to speak of) and when your service level is always below expectations. You must convince people to become your partners when all you have is belief in the future. And you must market your future ecosystem to investors when you do not even have a successful vertical application. Most importantly, you need to keep your faith in the ecosystem, and resist the temptation to take the quick but shortsighted path of doing everything yourself.

More than strategy, more than capital, more than luck, you need conviction. It will take a long time before you can enjoy all the nice things about a platform—critical mass, network effects, profit. Until then, you will just have to keep trying and believing.

Back to Top

The Weak Partner Paradox

Last year, Taobao produced about 12 million parcels per day. How do you handle a number like that? Our meteoric growth makes it impossible to run our own logistics operations: we would soon be a company with more than one million employees just running delivery centers. So we knew quite early on that we needed an open logistics platform.

But where to begin? By the time Amazon got its start, UPS, Fedex, and even Walmart had already developed mature business models, logistics networks, and human resources. You could easily build a team by hiring from these companies and leveraging third-party services providers. But China’s delivery infrastructure is weak, and its human capital even weaker. The country’s enormous size and varied terrain has ensured there are no logistics companies that can effectively service the entire country down to the village level.


We knew quite early on that we needed an open logistics platform.


So the first question I asked myself was: Can we build a fourth-party logistics platform when there have never been third-party service providers?

Therein lies the paradox. The strong third-party logistics companies did not believe in our platform dream, while the partners who wanted to work with us were either just startups or the weakest players in the industry. Obviously, the giants did not want to join us, considering us their future competitors. But could we realize our vision by working with startups who were willing to believe, but whose ability was really not up to snuff?

No growing platform can avoid this paradox. By definition, a new service, especially one that aims to become an enormous ecosystem, must grow from the periphery of an industry with only peripheral partners to rely on. At the same time, your partners will be fighting tooth and nail amongst themselves to capture bigger shares of your growing ecosystem. Your job is to guide them together toward a nebulous goal that everyone agrees with in a very abstract sense.

If you want to build a platform, take a good, hard look at your industry. Who can you work with? Who will work with you? Do they share your conviction? How will you help them grow as you grow with them, while at the same time supervising their competition and conflict? There are no easy answers to these questions. This is why we call it an "ecosystem": there is a strong sense of mutual dependence and coevolution. Over time, hopefully, you and your partners will become better and better at meeting your customers’ needs together.

This story has a happy ending. Our logistics platform handled five billion packages in 2013 (more than UPS), and now employs over 950,000 delivery personnel in 600 cities and 31 provinces. There are now powerful network effects at work: sellers’ products can be assigned to different centers, shipping routes can be rerouted based on geographical loading, costs and shipping timeframes continue to drop. Most importantly, we now have 14 strategic logistics partners. Once weak, they have grown alongside us, and are now professional, agile, and technologically adept.

Incidentally, one of Alibaba’s core ideals goes as follows. With a common belief, ordinary people can accomplish extraordinary things. Any successful platform begins ordinarily, even humbly, mostly because you have no other choice.

Back to Top

The Killer App Paradox

Back in 2008, when Salesforce.com was exploding and everyone started to realize the importance of SaaS (Software-as-a-Service), we deliberately tried to construct a platform to provide IT service for small and medium-sized enterprises in China. Taobao had already become quite successful, but China at the time had no good IT solutions for small businesses. We christened our new platform AliSoft, built all the infrastructure, invited lots of developers, and opened for business.

Unfortunately, AliSoft was a spectacular failure. We supplied all the resources a fledgling platform would need. However, there was one problem. Users were not joining, and the developers had no clients. Within two years, we had to change the business. The fundamental flaw was in our intricate but lifeless planning: we had a complete platform infrastructure that was unable to offer specific, deliverable customer value. There was no killer app, so to speak, to attract enough users that can sustain economic gains for our developers.

Alisoft taught us an important lesson. Platform managers cannot think in the abstract. Most successful platforms evolve over time from a very strong product that has profound customer value and huge market potential, from there expanding horizontally to support more and more verticals. You cannot design an intricate but empty skeleton that will provide a suite of wonderful services—this is a contradiction in terms.

To end users as well as partners, there is no such thing as a platform, per se; there is only a specific, individual service. So a platform needs one vertical application to act as an anchor in order to deliver value. Without that, there is no way you can grow, because nobody will use your service.

But herein lies the killer app paradox. If your vertical application does not win the marketplace, the platform cannot roll out to other adopters. And, making that one vertical very strong requires that most resources be used to support this particular service, rather than expanding the platform to support more verticals. But a platform must expand basic infrastructural services to support different verticals with different (and often conflicting) needs and problems. In other words, platform managers must balance reliance on a single vertical with the growth of basic infrastructure, which in all likelihood may weaken your commitment to continuing the success of your killer app.

What to do? How do you decide whom to prioritize when your business becomes complicated, when your ecosystem starts to have a life of its own? Managing the simultaneous evolution of verticals and infrastructure is the most challenging part of running a platform business. There is no magic bullet that will perfectly solve all your problems. You have to live through this balancing act yourself. Or put another way, your challenge is to constantly adjust on the fly but nevertheless emerge alive.

Some concluding words of wisdom for those who are not deterred by the long, difficult road ahead. Keep your convictions. Be patient. Trust and nurture your partners. And find good venture capitalists that can deal with you losing huge quantities of money for many years.

Back to Top

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More