Facebook's Portfolio Management
How Facebook navigated its business transition from desktop to mobile
This is an excerpt from Move Fast: How Facebook Builds Software, coming out July 6. Subscribe to this newsletter for advance content from the book.
“Facebook has been an incredibly innovative company because its engineering values system enables it to embrace risk.”
─Jocelyn Goldfein
The four critical years of Facebook’s transition from desktop to mobile occurred between 2010 and 2014.
Over these four years, Facebook explored several different business models as it searched for a new cash cow. Facebook’s revenue in 2010 came primarily from desktop advertising. In order to complete its pivot to mobile, Facebook would need to develop a new business model suited to a world where mobile usage dominated.
By the end of 2012, Facebook was starting to run mobile ads. At $73 million in annual revenue, the results of mobile advertising were promising, but not tremendous. Facebook was making almost the same amount of money from payments flowing through its desktop gaming platform. And both of these revenue sources were dwarfed by desktop advertising, the dominant source of Facebook’s $5 billion revenue in 2012.
At the same time, an hour north of Facebook campus in San Francisco, Ilya Sukhar was a few years into the development of Parse, a company he co-founded in 2011 to make cloud services for mobile developers. After studying computer science at Cornell, Ilya had worked at two different startups that had been acquired. With Parse, he was putting his startup experience to work with his own company.
Parse was a newer kind of cloud provider called “backend-as-a-service,” built to simplify the complexities of Amazon Web Services and the mobile ecosystem.
“Parse made cloud services for mobile developers,” says Ilya. “We wanted to make it dramatically easier to spin up an app and get all the basic functionality, particularly on the backend.” When Ilya started Parse, there was a strong desire in the developer world to build mobile apps. The rest of the world was realizing what Facebook had realized: the future of consumer technology was mobile.
To build a mobile app was a frustrating process. An application backend needed cloud storage, user accounts, analytics, push notifications, and all the other functionality needed to integrate comfortably with web, iOS, and Android frontends.
Parse gave developers a smooth experience for building mobile application backends, and developers loved it. That’s why Facebook decided to acquire the startup.
By 2012, Facebook had also become popular among developers. Game companies such as Zynga relied on Facebook’s desktop product for distribution and payments. The model was similar to that of the iOS App Store and the Android Play Store. Developers benefitted from various distribution and growth mechanisms, and in exchange Facebook kept a percentage of all payments flowing through the system. These payments were largely for virtual goods, a surprisingly large business in 2012, making up 15% of Facebook revenues.
But the team responsible for this business line, the Platform team, knew that this number was down from 19% the year before. And it would keep sliding downward as the majority of Facebook traffic swung from desktop to mobile. On the mobile platforms, Apple and Google played the role of distributor and payment facilitator. The platform team needed a new way to contribute to Facebook’s success in the mobile era.
Around the company, teams wondered: would mobile advertising be enough?
Mobile is a very different experience than desktop. It is a smaller screen, so developers considered that mobile users might consume fewer ads. In 2012, mobile e-commerce had not yet developed significantly, so maybe e-commerce ads would not perform well on mobile. And there was the potential that new mobile ad blocker technology would become popular. With all of these risk factors, Facebook was unsure if mobile ads would drive the same kind of revenues that the company was seeing on desktop.
Facebook was looking for alternative business models to supplement advertising, and was seriously considering a move towards cloud computing or some other kind of developer platform business.
The vision of Facebook as a developer platform made Parse an appealing acquisition target.
“Facebook had a little bit of an existential situation on their hands,” says Ilya. “The whole company had not yet done the transition to mobile. That was a big question in the stock market.”
Although Facebook was starting to make its technological shift to mobile, there were lots of unanswered questions about how mobile would make money for them. How would the ad units look? How should they be priced? How should Facebook’s sales and marketing teams work to improve those ads?
And most importantly--did advertisers even want to buy mobile ads from Facebook?
Ilya explains the predicament that caused Facebook to eventually buy Parse. “Facebook’s business model in a post-mobile world was undetermined. The core business was going to do fine for years, but it wasn't going to grow, because everything was transitioning to mobile.”
While Facebook was still navigating the open questions of how to monetize its mobile user base, the Platform team decided to try to explore cloud computing as a successor to its gaming payments business.
In April 2013, Facebook formally announced the acquisition of Parse. Ilya Sukhar describes the acquisition as one of several bets that Facebook was placing to answer the question of Facebook post-mobile. “We were one of the experiments.”
“When we sold the company to Facebook, we had somewhere between 60,000 and 80,000 active developers. Through our height at Facebook, it was a magnitude higher than that. We had all these developers, we had data centers, we had technology that was really interesting.”
Facebook’s vision was to use Parse as the foundation for a developer platform that could unify mobile and backend infrastructure together with Facebook’s social identity platform. It was a bright future, but it was never to be realized: mobile advertising turned out to be far more lucrative than Facebook anticipated.
By the end of October 2013, just six months after Facebook announced the acquisition of Parse, Facebook mobile ad revenue had reached parity with desktop ad revenue. By the end of the year, mobile ad revenue was larger than desktop. And the Platform team had run another experiment: creating an ad unit that allowed developers to pay for installs of their mobile apps. It was a huge success within months. Across the company, the anxiety around the viability of mobile advertising melted away.
Online advertising businesses are notoriously hard to build, but when they are successful they generate more cash than almost any other Internet business model, including cloud infrastructure. Once it was clear that mobile advertising would be a powerful business for Facebook, the company centralized all of its spare resources on building the cash cow.
With the success of mobile ads, Parse fell by the wayside.
Ilya sighs. At the time he felt conflicted: his acquirer’s stock would now be worth much more, but his company had to be subsumed in the process. “No matter how well Parse did, no matter how much developer love we had, how many apps we had, how many users of those apps we had, how much revenue we were making -- at the end of the day, it's really hard to compete with a well-oiled ad business. Ad businesses are very efficient. Very, very efficient.”
Parse found itself under the control of an owner that now had no reason to invest in a cloud infrastructure platform. Facebook had crossed the chasm into the post-mobile world. “The company had made the transition. Stock was up a ton, printing money, newsfeed working great, newsfeed ads working great, app install ads working great for the platform team.”
As engineering headcount was increasingly allocated to teams working on mobile and advertising infrastructure, Parse’s resources dwindled.
Ilya is sanguine about the experience. “An engineer put toward ad targeting, put toward the core business at that time was just a way better investment for Facebook. I can't really argue with that. I think the most fundamental explanation is that Facebook didn't really need Parse. Fundamentally, the company moved on. It's hard to say exactly when, but I think the Parse acquisition somewhat became irrelevant over the time we were there.”
Ilya was not happy about the situation. His own business had become a casualty of the acquirer’s success.
“It hurt a lot. At the time it was just brutally painful. If you had lopped us off, just cut us off and made us an independent company again, I think we would be a very strong startup.”
Ilya expresses respect for Facebook. Mobile ads are a huge cash cow, with much higher expected value than cloud services divisions. Facebook had shifted its mentality from a hedging strategy to one of complete focus on mobile advertising. Parse was no longer a hedge--it had become a distraction.
Ilya gives his high-level conclusions of what makes Facebook a successful company. “Facebook made a decision to try things. If they didn't take off quickly, those things were abandoned.”
“When I was at Facebook, the company’s core strength seemed to be Mark Zuckerberg’s management orientation. Every six months he had a few priorities. He oriented the company around those priorities. Teams got shuffled. Sometimes, messes were left behind. But we were always super focused on the strategic priorities. We were intellectually honest about where the leverage was in the system. I think there are a lot of advantages to that. It worked. It's hard to argue with.”