The land where Facebook loses money
This post is based on a report by Caribou Digital, with support from Mozilla Foundation. You can download the full report from our website.
Actually, Facebook makes a lot of money. More than $8 billion in just the last quarter of 2016, almost all of it (~98%) from advertising. Those revenues have helped Facebook become one of the largest firms in the world (currently #7 by market cap), and it, along with Google, so dominates the digital advertising industry that some analysts estimate 85% of all new ad spend in the U.S. will go to the duopoly.
While most of the money that Facebook earns comes from the developed economies of North America, Europe, and East Asia, user adoption in these regions has plateaued. To satisfy shareholder expectations for continued growth, Facebook (and other global platform firms) are looking to the emerging economies of South Asia, Latin America, and Sub-Saharan Africa as the growth markets of the future,  and you are just as likely to hear about “bringing the next billion online” on a Facebook investor call as at an international development conference.
The problem, however, is that it’s hard to make money in these markets. By our estimates, Facebook is probably losing money in many of the low-income countries in which it operates, especially in Africa and Southeast Asia. There are two main reasons:
- Advertisers don’t pay very much to show their ads to poor people. Low disposable incomes mean less consumption, and therefore fewer segments where advertising shows an ROI.
- Poor-quality devices, unreliable networks, limited digital literacy, and expensive mobile data limit how most people in emerging markets engage with digital content and services. One consequence is that showing them lots of targeted, high-value ads is hard.
Ironically, Facebook has done as much or more than any other company to tailor its product to these populations and market environments, and it clearly believes that bringing more people online is both a social good and benefit to its bottom line. But doing so has required heavy investment in alternative products: It developed a Java-based version of Facebook, for feature phones. And a “lite” version of its Android app for low-budget smartphones. And the zero-rated (no data costs) version that many operators offer to entice users, aka Free Basics. These alternatives enable more people to sign up and use the product, but the user interface and bandwidth restrictions mean Facebook can only serve fewer, lower-value ads (or in the case of Free Basics, no ads at all), further limiting revenues.
The end result is that there are hundreds of millions of Facebook users who don’t really generate much revenue for the firm, but for whom Facebook still has to bear the operating costs.  And the fact that Facebook has created multiple versions of its product, probably with largely distinct code bases and developer teams, means it takes a hit on economies of scale as well. It’s true that emerging market users consume much less data and bandwidth, which lowers hosting costs for Facebook compared to a U.S. or European user. But some of that savings may be offset for users of the Java version and Android Lite version, both of which rely on sophisticated proxy architectures to offload processing and storage to the cloud.
To better understand the geography of Facebook’s revenue, we conducted a country-level analysis using a combination of data from Facebook’s own ad tool, public filings, and the results of our own ad campaign. Following research that has shown the correlation between G DP and advertising revenues, we first tested and confirmed the relationship between advertising demand (CPC rates) and GDP per capita across a sample of both developed and developing countries. We then weighted the regional ARPU figures published by Facebook by the GDP/capita results, giving us country-level ARPU figures. Combined with MAU estimates based on scraping Facebook’s ad targeting tool, we created rough estimates of total revenue and costs (using MAUs as imperfect proxy) per country, as shown in the chart below.
The outsized role of the U.S. is immediately clear — it provides almost 50% of all Facebook’s revenue, about $4 billion, with only 12% of the user base. And it’s also highly saturated, with 69% penetration in Facebook’s home market. Given that it’s such an outlier, excluding the U.S. from the chart allows us to see the main cluster of markets in much better detail (below).
Here we can see that, predictably, the large Western and East Asian markets (excluding China, where Facebook is blocked) are tops in terms of total revenue. Brazil also does well, by virtue of a huge Facebook user population (estimated at 125 million) and relatively large middle class that attracts advertising dollars. The challenge for Facebook is those countries that are down and to the right — lots of users, but little revenue. India, Indonesia, Philippines, Vietnam, Bangladesh, Nigeria, and others have millions of users, but their very low ARPUs keep total revenues low.
Without actual cost information we can’t know with certainty which markets are or are not profitable, but somewhere in this chart there is a positively sloping line that divides the profitable (above the line) from the unprofitable (below the line). At an operating cost per user of around $1 per quarter (shown as an orange line), dozens of countries — including larger ones such as Nigeria, Indonesia, India, Vietnam, and the Philippines — would likely be unprofitable. If that line is closer to $0.50 per quarter, most of those would barely break even, and some countries — notably, India — still might not be making money.
Why bother? Facebook clearly believes in the long-term potential of some emerging markets, and in places such as India, where it has very publicly expressed its commitment to build up its business, the long-term prospects do seem favorable. India has a rapidly growing middle class (an estimated 380 million Indians will join the middle class by 2022), strong government support for digital infrastructure and business, and a highly competitive telecommunications sector bringing down prices for consumers (especially since the entrance of Reliance Jio last year), all of which bode well for Facebook’s core business.
And Facebook isn’t just waiting for these ad markets to grow. The company is making ambitious investments into next-generation Internet access technologies (e.g., the Aquila drones) as well as more conventional approaches (e.g., Express WiFi in Kenya and India) in an effort to make access more affordable for a larger segment of the population. Bringing down data prices (vs. extending network coverage into rural areas) helps to address the affordability constraint that limits user access and thus advertising revenues. Facebook isn’t the only global platform looking to integrate vertically — Google is building fiber in Africa, Wifi hotspots in India, and has its own ideas for rural network connectivity (balloons). For advertising businesses, bringing more eyeballs to more screens, for longer and richer engagements, drives revenue.
Facebook is also moving more decisively into financial services. It’s expanding its payments functionality in Messenger (e.g., group payments, international payments via TransferWise, social commerce via Qwik), and is launching payments within WhatsApp in India using the UPI rails. When it initially launched payments within Messenger in 2015, the company said it wasn’t setting out to build a “payments business” that monetizes directly via transactions, and indeed the more natural fit for the company would be to keep its payment products restricted in function and dollar amount (to avoid higher levels of regulation), and keep everything free for end-users as a way to build engagement.
The obvious comparison here is Tencent’s WeChat, which has long transcended its role as a messaging app to become the do-everything platform in China, with convenient payments at the core. With 900 million MAU and a stickiness that surpasses Facebook (daily averages of 66 min. vs. 50 min), WeChat has demonstrated the attractiveness of a unified experience that allows users to make payments, play games, send messages, and more, a model that runs contrary to the silo’ed, discrete consumer internet platforms of the West (i.e. separate apps for each function). If Facebook does start to emulate parts of WeChat’s model, it will be interesting to see whether the firm does try to diversify with a more transactional revenue model (Tencent earns only ~20% of revenue from advertising), or maintains itself as a pure advertising business.
Facebook can afford to operate in low-income markets because it has a firehose of profits pouring in from markets in North America and Western Europe, revenue that it can use to cross-subsidize its operations in Nigeria or Indonesia. On one hand, this is a fantastic deal for users in Nigeria, Indonesia, and other lower-income countries — they are getting a sophisticated, world-class digital product for free, even if their data and activity aren’t worth much to the company.
But for local or regional firms that may want to compete with Facebook, this presents an almost insurmountable problem. Facebook is free, so if you want to run a social network, or a messaging app, or maybe now P2P payments, you also have to be free, which probably means monetizing via ads. But unless you are also monetizing in the lucrative markets of the West and East Asia, there probably isn’t enough ad revenue in your domestic market to support a competitive product. The result is Facebook has almost unassailable moats against local/regional competitors in emerging markets.
The implication for internet-based businesses in emerging markets is that instead of the Silicon Valley Internet business model, where moving up the user adoption hockey stick is the success metric that keeps the venture capital flowing, Global South businesses have to build more strategically, working toward profitability from the beginning. Not having the ability to pull the advertising trigger and instantly start monetizing their user base may therefore be a helpful constraint, as it forces these businesses to instead develop more disciplined revenue models that can support more organic growth.
 ”Media Websites Battle Faltering Ad Revenue and Traffic.” 17 Apr. 2016,
 By our estimates, there are about 400 million users across Africa, Latin America, and South Asia using F4EP, Android Lite, or Free Basics: Facebook announced in July 2013 that F4EP had reached 100 million users, and based on its fan page, we estimate it has at least 200 million in 2017 (users are encouraged to “like” the fan page when downloading the app; public numbers from the company show the number of “Likes” (currently ~500 million) have historically been roughly 100% to 130% higher than the actual number of users). Per the company, Facebook Lite has already reached 200 million users in less than two years, while Free Basics has a reported 25 million users.
 The macro-economic research clearly shows strong correlations between a country’s GDP and ad industry spend, a trend that has persisted for most of the last century.
 Facebook’s most recent 10-K shows a cost of revenue that, divided by monthly active users, averages out to a $0.56/user per quarter, but this doesn’t include the relevant costs from its tremendous data center infrastructure capital investments. One point of comparison might be Snap, which uses Google’s cloud for all its hosting and processing needs (and therefore reports those costs explicitly); the company paid $0.60/user per quarter in hosting costs. XXX
 For a detailed discussion on the trade-offs in increasing mobile internet access, including extending network coverage vs. increasing speeds or lowering costs, see Caribou Digital’s report on Digital Access in Africa.
 WeChat (along with rival payment platform Alipay) has become so dominant that cash is disappearing in China’s cities, and the two platforms are set to surpass Visa and MasterCard in global daily transaction volume. https://www.nytimes.com/2017/07/16/business/china-cash-smartphone-payments.html?mcubz=0