2.

Foster the Full Stack

Don’t Just Rely Only on Software

In Silicon Valley’s classic model, startups are “asset light.” They focus on one piece of the value chain and solve a customer’s problem using an innovative product or solution—one whose launch ideally requires limited capital, hardware, or complexity.

If you want to build an audacious, world-changing photo-sharing platform, then this strategy works effectively. The existing technology and infrastructure ecosystem allow you to focus on the user interface and partner with others for storage, user authentication, and social media integration. Similarly, even for complicated ventures like on-demand delivery, many parts of the value chain—such as street addresses, local maps, route optimization software, and logistics support—can be obtained from existing technology or ecosystem providers.

Silicon Valley’s asset-light modus operandi is enabled by a rich tapestry of companies that provide the necessary infrastructure for disruptors. Other dogmatic startup principles—such as the imperative to create a lean startup—further encourage entrepreneurs to focus narrowly on one part of the value chain and one product. The idea is to build excellence in one area and get the rest from the ecosystem.

Yet for most of the world, and particularly for Creators no matter where they are located, this model is simply impractical. Because the rich ecosystem of enabling technologies is typically nonexistent or is severely lacking for the type of business they are building, Frontier Innovators often need to construct the “vertical” full stack themselves—developing both the ultimate product or service and the enabling infrastructure that underpins it.

The full stack isn’t always a vertical one. Often startups build the “horizontal” full stack—offering a wider range of products and services than their Silicon Valley equivalents in order to create an ecosystem.

Building the Vertical Stack

Two men who exemplify building the vertical stack are Ben Gleason and Thiago Alvarez. In 2012, they decided to tackle the critical need for financial inclusion in Brazil. This South American country has some of the highest interest rates and balances in the world: the average bank lending rate is greater than 50 percent (compared with 5.5 percent in the United States and 1.75 percent in the United Kingdom).¹ At the same time, Brazilians have high levels of short-term credit card debt.² Many customers do not understand their financial situation holistically, and they lack the tools that would empower them to take concrete steps to improve it.

In the United States, Mint created a digital personal finance manager (PFM) to address a similar challenge. The product quickly attracted more than 1.5 million users and was ultimately sold to Intuit.³ Now US customers have their choice of a panoply of PFMs, ranging from apps to free products provided by banks. Specialized service companies like Credit Karma, which helps customers improve their credit scores, focus on improving key aspects of the customer’s financial life.

Brazil had no such plethora of options. To solve this challenge, in 2012 Ben and Thiago launched Guiabolso, Brazil’s first PFM. Below the surface, Guiabolso looked very different from its rapidly scaling US counterparts (Mint and Credit Karma).

At first blush, Guiabolso appeared to be a budgeting tool. It allowed consumers to enter their spending habits to track how much money they had used and how much would remain at the end of the month. But like many apps where customers self-reported spending, Guiabolso faced the GIGO challenge: garbage in, garbage out. The app’s insights were based on what customers input. Unless customers reported information accurately and consistently to the app, the insights the app provided would be worthless.

For Silicon Valley PFMs, resolving this was an easy challenge. Mint gets actual financial data straight from customer bank accounts, bypassing the customer-input phase. Mint can do this because preexisting platforms like Yodlee connect the PFM to banks.

These platforms did not exist in Brazil in 2012. Ben and Thiago had to decide: would they continue with the imperfect budgeting tool, or bet the fragile startup’s future on building the bank connection layer themselves?

They decided to bet. Developing the platform was no easy feat. Brazil’s banks have excruciatingly variable security features. It took more than a year to create a stable platform that could reliably access banking data. And that was on top of the need to make sense of the complex data, create insights from the information, build an engaging customer front end, and develop a strategy to reliably find users and get them to join the platform.

As soon as Guiabolso had solved that issue, the next problem emerged.

To engage users around their biggest pain point—credit card debt—the app needed to give customers insights into their creditworthiness and their ability to access lower-priced credit. Brazil did not have an all-encompassing FICO score equivalent. For most Brazilians, the credit score is binary: you are either in default or you’re not.⁴ That is not very insightful for most customers. Nor is it helpful for banks that make credit decisions on customers who are not their clients.

So Guiabolso built a proprietary Financial Health Index. In addition to informing customers whether or not they are on the blacklist, the app proactively gives them an impartial view of their financial health and sets them up with the tools to improve it.

Guiabolso’s customers now understood their current financial health and learned ways to improve their creditworthiness. Naturally, next they wanted to benefit from these insights. In the United States, Mint and Credit Karma helped customers secure more-affordable loans by matching them with providers across the country. Guiabolso thought it might be able to do something similar.

Again, however, the company was stymied. The traditional lenders in Brazil were reticent to provide targeted offers through the platform. Many were uncomfortable with this new method of acquiring and serving customers online, particularly because they were unaccustomed to issuing fairly priced loans on the basis of a credit score to customers with whom they had no direct relationship. Unlike the United States, with its rich ecosystem of financial technology (fintech) lenders, Brazil had none.

Guiabolso had to build the infrastructure itself. It launched a de novo financial lender in Brazil that would lend through Guiabolso’s platform. It opened the platform to up-and-coming lenders and banks that wanted to increase their loan book and were eager to reach new customers, and yet weren’t equipped to sell loans in a rapidly moving digital format. In this way, Guiabolso’s product is transformative, offering users access to credit at lower rates than rates Brazilian consumers can otherwise obtain in the market.

Guiabolso has more than five million users on the PFM and has distributed more than $200 million in loans. It has also raised more than $80 million in venture capital funding (including from me when I worked at Omidyar Network).⁵

To achieve this traction, Guiabolso had to create four separate businesses in order to provide a single product: an interconnection layer to connect to banks digitally; a consumer app with valuable insights for customers; the Financial Health Index and credit-scoring platform; and a lending product to jumpstart the marketplace. It was no easy feat in any market but particularly challenging in Brazil.⁶

Guiabolso may seem to be an exceptional case. Yet many of the Frontier Innovators you’ve met so far have faced a similar challenge. If anything, Guiabolso had it easy. Many Frontier Innovators must build physical enabling infrastructure, rather than only software. To offer their home energy systems, for example, Zola (see the introduction) built an R&D facility to design low-cost appliances, a rural sales force and centralized call-center operations, a customized software solution for its sales force, and an in-house financing business to offer customers a pay-as-you-go service. Similarly, M-PESA could not build a fully digital payment network. Instead, it had to cultivate a network of mom-and-pop stores across the country to allow customers to deposit and withdraw cash from the system.

Inanc Balci, a co-founder of Lazada Group, a leading Southeast Asia e-commerce platform, summarized the dynamic: “It is absolutely counterintuitive that to launch an e-commerce business, we had to create an entire logistics business. But for us, it was consequential for survival.”⁷

The stories of Guiabolso, Zola, and Lazada naturally raise questions about when it makes sense to build the full stack and how entrepreneurs should prioritize precious resources.

When to Build the Full Stack

The decision to build more pieces of the stack should not be made lightly. It requires capital and time—both critical resources for startups. It also increases interdependencies and risk, because the pieces of the stack must work together seamlessly.

To decide when and what to integrate, an innovator should begin by considering a series of questions.

Does the Ecosystem Have the Infrastructure You Need?

The process begins by understanding what is required for your business model to function and whether others in the ecosystem provide the requisite pieces at a sufficiently high level of service.

Guiabolso required two linchpins. The first was a front-end customer application or web experience that provided insights about how customers used their money and where they could improve. This was core to the business, and something that the company would build. The second key ingredient was the bank interconnections to enable customer data to be fed automatically to the app. Guiabolso would have preferred not to build this layer. It conducted a market study before determining that no suitable solution existed.

Some businesses decide to vertically integrate more parts of the value chain than strictly necessary (e.g., Apple and Tesla, which vertically integrate the product design, distribution, and large parts of the supply chain of their products), deeming this model core to their competitive advantage. But innovators in developing ecosystems often find that the vertical full stack is a matter of necessity, building essential infrastructure just to make the core model work.

Frontier Innovators should evaluate any alternatives that exist in the market. The available infrastructure will differ in various regions and industries. Startups located in the United States but outside Silicon Valley—say, in Detroit or Milwaukee—will still benefit from the rich US technological infrastructure. An entrepreneur building Lazada in Detroit could tap in to the established US payment and shipping ecosystems. Similarly, it would be redundant for a PFM startup in Europe to build a new credit-scoring infrastructure.

Ultimately, if innovators can borrow a stack, or if it is available in a commoditized manner at a sufficient quality level, they do not need to build it and should not waste resources. If they cannot use an existing stack, then they must explore the next question.

Can Others Provide the Stack?

Sometimes, just because the stack is not available doesn’t mean it can’t be accessed. Frontier Innovators can enable others to provide the stack for them, often through technology. Sacha Poignonnec is the co-CEO of Jumia, a leading pan-Africa e-commerce platform and the first technology company in Africa to go public on the New York Stock Exchange. Sacha emphasizes that certain elements can be outsourced if the right enabling tools are provided. When Jumia started scaling its e-commerce platform, no one provided deliveries with the requisite level of service in Jumia’s markets. To offer end-to-end logistics in rural African markets—with item tracking, delivery time estimates, and other features now standard in the West—Jumia offered technology tools for local entrepreneurs to track delivery, optimize routes, and collect payments. Jumia created these enablement tools internally but empowered others to fill the gaps. Now Jumia has more than eighty-one thousand partners on its system in fourteen African markets, including Egypt, Morocco, Uganda, and Nigeria.⁸

An important reflection is whether partners can scale with you. David Vélez, CEO of Nubank, learned this the hard way. Nubank is a credit card company for the mass market, operating in Argentina, Brazil, and Mexico. David explains:

When we started, we worked with a third-party credit card processor, but realized rapidly their technology was not strong enough to scale. We also worked with a banking partner because we didn’t have a license initially. However, there was too much risk depending on them, so we had to get our own. We faced the same problem with customer service. We decided to build it in house from the beginning, rather than depend on a third party. It was the only way we could guarantee quality.⁹

For Nubank, local solutions existed but were not strong enough to scale with the business. David had to build each in turn. Nubank has raised more than $800 million and now is a leading Latin American fintech worth more than $10 billion.¹⁰

Can You Stage the Full Stack Over Time?

Building a startup is extremely challenging. Having to build a vertical stack—the equivalent of building multiple startups at the same time—often proves impossible. Some pieces of the vertical stack must be built in tandem, but often not everything. Staging is key.

For Guiabolso, the PFM was the first crucial product. This meant the company had to build the bank interconnections immediately. The rest, however, could wait. If customers didn’t care about the PFM insights or if the bank interconnections didn’t work, it wouldn’t matter if the company could build a lending platform. That’s why Ben and Thiago decided to wait a few years before providing the credit-scoring and lending products.

Deferring might mean others will build instead. Markets are dynamic, and others are constantly filling out pieces of the stack. Companies like OkHi are building different pieces of the infrastructure in their areas, including street addresses and logistics. Similarly, platforms like M-PESA and its many replicators around the world are creating the payment ecosystem. Before investing in vertical infrastructure, you will find it worth investigating whether other pieces of the stack may get filled out with time.

Ultimately, for Frontier Innovators, having to build the full(er) stack is not a strategic choice; it’s a practical reality. Indeed, building the full stack is one of many examples in this book where Frontier Innovators have to do more with less. If at all possible, finding partners or enabling others to support you are valuable options to explore.

This may sound daunting. However, there is a bright side. Building the full(er) stack confers competitive advantages as well.

The Full-Stack Moats

Building the full stack is a double-edged sword. Although at first it makes the mountain that much higher to climb, it also makes it that much harder for anyone else to imitate the model afterward. Frontier Innovators often enjoy three types of full-stack moats (a natural defense that allows a company to maintain its advantage over time): competitive, capital, and technical.

The Competitive Moat

The competitive moat is clearest. In Brazil, for a competitor to create a competitive PFM and a credit-scoring lending product would require rebuilding much of Guiabolso’s infrastructure. It’s no wonder Guiabolso doesn’t have much competition, even after five years. Those that move first and get a head start in launching gain significant advantage, because once competitors notice the opportunity, they are still years away from being able to replicate the model.

The Capital Moat

This same phenomenon drives a capital moat. Guiabolso has raised more than $80 million.¹¹ For other companies looking at the space, the knowledge that even replicating the existing offering would take a similar amount of capital is daunting. Funding such a competitor is unappealing to investors. Venture capitalists would consider it a losing proposition to invest in the number two player that is a number of years behind (particularly in markets that are dominated by network effects and where one winner will emerge). Saed Nashef, co-founder of Sadara Ventures, which invests in capital-starved Palestine, explains: “Capital can be a powerful advantage. When companies raise large rounds ahead of competitors, it becomes a driver to become a winner.”¹²

This capital moat plays out in another, more nuanced way. In Brazil, for example, there are limited investors who lead large rounds. By its third funding round, Guiabolso had already received capital from many of the leaders: Kaszek Ventures (one of Latin America’s leading early-stage funds, which is covered in chapter 11); the International Finance Corporation (IFC; the private investment arm of the World Bank); Omidyar Network (my former employer, a global philanthropic investment firm); Ribbit Capital (a global fintech investor); and Valor Capital (a cross-border Brazil venture capital fund). Guiabolso was able to do this before another competitor emerged in a meaningful way.¹³

Investors are often reluctant to support two competing companies in the same market. Because being full stack often requires more capital, it is an opportunity to lock up more of the leading investors in the space, thus making it more challenging for an up-and-coming competitor to thrive.

The Technical Moat

Being full stack also confers a technical advantage. As Guiabolso built its interconnection capabilities with banks, many of them were opposed to the company’s central idea. It is within a customer’s rights to share their read-only internet banking credentials and allow Guiabolso to access their account information. However, many banks resisted, rightly believing it would decrease their relationship with the customer. Banks developed increasingly sophisticated ways of thwarting Guiabolso’s attempts at reading customer data, but Guiabolso became increasingly proficient at getting through. New competitors coming in would have to repeat the technical and strategic iterations that Guiabolso went through.

Similarly, Fetchr, which you will get to know in chapter 4 and which offers last-mile deliveries in the Middle East, had to integrate into a range of e-commerce platforms; in doing so, it learned what it takes to fix bugs rapidly to provide customers a seamless experience. Developing this kind of expertise takes many iterations and discourages online vendors from experimenting with other providers.

This is the full-stack moat. Frontier Innovators often endure significantly more organizational complexity upfront than their Silicon Valley counterparts. However, this yields a better position for them to endure and succeed over the longer term.

Building the Horizontal Stack

So far, we have focused primarily on the vertical stack—the infrastructure required to enable primary business functions. Often, entrepreneurs need to go one step further. To bring customers into the system, they have to offer a wider range of products than their Silicon Valley equivalents, either directly or via partnerships. Frontier Innovators do this much earlier in their journeys than conventional wisdom recommends. This is the horizontal stack.

Gojek: The Full Ride

The traffic in Jakarta is among the worst in the world. It often takes well over two hours to get from one end of the city to the other. And it’s only getting worse. Over the past decade, the city grew to more than thirty million people, and five million cars and fifteen million motorbikes clog the streets.¹⁴ Car ownership is increasing nearly 9 percent every year.¹⁵ Fewer than 25 percent of Indonesians use public transport.¹⁶

Ojeks, low-cost motorbike taxis, represent one way around this problem. While these motorbikes are less comfortable than cars—subjecting riders to frequent rain, pollution, road noise, and higher accident risk—they do something cars can’t. They weave through traffic during the frequent congestion. If you want to arrive in style, you take a car. But if you’d like to arrive on time (and affordably), you take an ojek.

In 2010, the ojek market was informal and unregulated. Many ojek drivers operated with little documentation, few safety standards, and little tracking. Drivers had difficulty locating would-be passengers. They often took home fewer than three or four fares a day, spending hours driving around (or waiting in traffic), hoping to get a passenger.¹⁷ Demand couldn’t find the supply, something that consistently limited the market.

Nadiem Makarim wanted to do something about it. After graduating from Harvard Business School in 2011, he started a centralized ojek taxi service: a physical call center that identified drivers and directed them to customers. Nadiem started with six personal phones, twenty drivers, and customers from his personal network. The business solved a personal problem for him and his family but did not expand beyond sustainable subscale operation for a few years.¹⁸

In January 2015, Nadiem finally secured the funding to infuse technology into his small call center, and he launched Gojek. His vision was to formalize the entire informal ojek economy through a centralized technology platform. Gojek operates similarly to the Uber app, allowing users to call an ojek wherever and whenever.

For drivers, Gojek’s value proposition was clear. Historically, they worked long hours but completed only a few rides every day. With Gojek, they could easily double their orders. Although prices are lower, drivers make more money (and with more consistency) due to the reliable order flow. Furthermore, Gojek gives every driver on the platform a helmet (often for the first time), a uniform, and access to a high-quality motorbike.

Gojek also proved valuable for customers, allowing them to secure an ojek when they needed one (it’s important to avoid waiting in the rainy season) and quickly travel to any location in the city. Customers also found a safer alternative to the informal system in Gojek. Nadiem says it’s one of the reasons the majority of Gojek’s riders are women.¹⁹ Much like other ride-hailing businesses, Gojek offers a fixed price, eliminating the need to haggle with the driver.²⁰

By the end of Gojek’s first month, there were one thousand motorcyclists on the app. By the end of the year, the company had two hundred thousand drivers in five cities and more than four hundred thousand downloads. Gojek’s popularity skyrocketed, its brand aided by the suddenly ubiquitous green helmets and jackets peppering the city.²¹

More Products, More Services

Yet to truly scale, Nadiem knew he’d have to offer more than transportation. Many of his customers were unbanked and did not have access to credit cards or other digital means of payment, so they needed to pay for rides in cash. In Indonesia, 48 percent of the population doesn’t have a smart phone, and 33 percent have only a simple-feature phone.²² Gojek was an early online experience for many customers, making the platform an ideal place to offer a range of services. Furthermore, many drivers had free time outside commuting hours, when demand for transport tended to spike. To bring more customers onto the platform (and often online for the first time), and to engage drivers throughout the day, Gojek created an entire ecosystem of products and services.

The first element was support for digital payments. Most Indonesians were unbanked or underbanked. Yet Nadiem knew that digital payment would facilitate product adoption and greatly improve the customer experience.²³ Furthermore, digital payments represented an opportunity to create a significantly wider ecosystem.

Inspired by the success of platforms like Tencent’s WeChat and Alibaba’s Alipay, which have 1.1 billion and 900 million users, respectively, Nadiem set out to make Gojek the leading payment platform in Indonesia. Gojek already had two critical elements: an agent network (drivers) to get cash into and out of the system, and a ready transaction ecosystem (rides).²⁴ The latter is particularly critical because it is habit-forming behavior and creates a repeatable use case. In 2017, Gojek acquired Mapan, a community savings network; Kartuku, an offline payment-processing company; and Midtrans, a payment gateway, to build the network further.²⁵

Now Gojek drivers can also offer financial products and services—a huge step beyond traditional ride-sharing services. Drivers act as human ATMs: riders deposit and withdraw money from GoPay, Gojek’s mobile payment ecosystem, directly through drivers. Customers can then use GoPay to make payments (e.g., paying bills or transferring money to other Gojek users) and accumulate savings (by keeping a balance in the account). Over time, a greater range of financial products will be offered, including insurance and loans.

Gojek went even further. To respond to customers’ need for a wider selection of products in one destination, and to smooth demand for its drivers, the company’s range of services now includes food delivery (GoFood), commerce (GoMart, GoShop), massages (GoMassage), shipping (GoSend), and cell phone airtime (GoPulsa). One day, Nadiem hopes customers will be able to get almost anything from Gojek’s singular platform.

Nadiem aims to create a full ecosystem. As he explains it, “In the mornings, we drive people from home to work. At lunch, we deliver them meals to the office. In the late afternoon, we drive people back home. In the evenings, we deliver ingredients and meals. And in between all this, we deliver e-commerce, financial services, and other services.”²⁶

The strategy is working. By mid 2019, Gojek had two million drivers, three hundred thousand merchants, and one hundred million monthly transactions on its platform.²⁷ Major players like Tencent, JD.com, KKR, and Sequoia Capital back the company. In February 2018, Gojek closed a Series E financing round of $1.5 billion. Tencent led the round, with participation from investment firms across the United States, China, Europe, and Indonesia.²⁸

Nadiem, however, is proudest of a different accomplishment. As he explained, Gojek has become the largest income source in Indonesia. Millions of people receive income through Gojek. Nadiem emphasizes that for him the most important part of this victory is that Gojek has “increased the opportunities of [its] drivers.”²⁹

Approaching the Horizontal Stack

Building the horizontal stack is a different manifestation of the full(er) stack approach. Whereas entrepreneurs like Ben Gleason and Thiago Alvarez had to build enabling infrastructure only to offer their ultimate product, Nadiem Makarim had to build horizontally, offering related business models that mutually reinforced the core offering.

For many Frontier Innovators, building an ecosystem of activities is a strategy to create a viable business. As with the vertical stack, the horizontal stack should start with an examination of which pieces of the ecosystem are required, and in what order. For Gojek, this started by creating a core ride offering and rapidly expanding to other product categories. Vijay Shekhar Sharma, the founder of Paytm, a venerable Indian technology company worth more than $15 billion, explains: “Eventually businesses become an ecosystem at scale. In emerging markets you have to do it earlier to monetize within your customer base. At Paytm, our ecosystem includes payments, of course, but online commerce, a digital travel agency, [and] wealth management, among others.”³⁰

In some cases, building the horizontal stack may not be strictly necessary. Rather, it offers an opportunity to leverage an existing customer base to capture more of a nascent market, building the profitability of the business and improving its competitive positioning at the same time. Gojek has an opportunity in Indonesia to go much further than ride-hailing apps in developed markets. Rather than being constrained to a narrow product category (e.g., Uber and Lyft for rides, Instacart for deliveries), Gojek has a viable chance to create a “super-app,” which would allow it to dominate a number of sectors. WeChat has taken a similar strategy in China, in some ways becoming the de facto platform for consumers in China to access products and services on the internet.³¹

Often, Frontier Innovators can enable others to provide the horizontal stack with them. Gojek partners with others for most services (e.g., restaurants, merchants, and masseuses). Similarly, M-PESA enables anyone to leverage its payment platform, including financial institutions that offer services spanning insurance, savings, and loans.

Sometimes, however, an innovator needs to kick-start a nascent ecosystem by building the first pieces internally. Ctrip, the Chinese travel company, started as a travel community review service similar to Trip-Advisor. The challenge was that even if the online portal existed, it was challenging to book hotels or purchase flights online. As a first step, Ctrip methodologically partnered with existing hotels and airlines in Shanghai and other major Chinese cities to add them to the platform. However, the inventory of flights to the interior of China was limited, and once tourists arrived, there were even fewer hotels and organized local tourist activities. Ctrip itself played a role in creating the Chinese travel ecosystem, launching vacation packages with charter flights to select destinations, building hotels, and organizing tours.³² Over time, the ecosystem flourished. Now the company has a market cap of more than $24 billion, and more than 250 million users.³³

Ecosystems often have powerful network effects. As others provide services within the ecosystem, the innovator’s centrality to the model will only get stronger. As Gojek enables more service providers to sell through the Gojek platform, or as more businesses use M-PESA to offer financial products, Gojek becomes the de facto standard for customers. Innovators driving these network effects, like M-PESA, Gojek, and Ctrip, enjoy an enduring advantage.

Putting It All Together

In Silicon Valley, best practice suggests providing a software-based, asset-light product or service. Startups are told to focus on their “wedge”—the narrow segment in which they will compete and hopefully dominate. Providing a full end-to-end experience is a bold, expensive, and often foolhardy strategy.

Indeed, operating as a full stack seems intimidating. It confers more operational complexity and risk and increases the likely time to reach scale. However, the strategy also has distinct advantages, particularly since it offers competitive, capital, and technological moats. Similarly, by building a horizontal stack, Frontier Innovators can capture a greater share of the market, increase their relevance to customers, and improve their competitive long-term positioning.

Full-stack approaches are not new. They are reminiscent of Silicon Valley twenty years ago. To launch a business, many startups had to build their own servers in house. To create new products, they hard-coded—literally. Innovators were not only building their products or services but also creating enabling infrastructure and ecosystems.

Marc Andreessen, founder of Netscape and now a leading Silicon Valley venture capitalist, captured Silicon Valley’s zeitgeist when he proclaimed, “Software is eating the world.”³⁴ This does appear to be true in Silicon Valley. Yet for entrepreneurs operating in nascent ecosystems, and especially for Creators, software is rarely sufficient to solve complex problems or build new industries. Perhaps “Software-enabled technological solutions, incorporating a more expansive view of the stack, are eating the world” is a more accurate, if less catchy, slogan for the Frontier.