Facebook, the start-up that collected start-ups?

Facebook’s acquisition of Whatsapp shows that the self-styled disruptors are themselves living in a baffling environment where old certainties are of little value. Facebook itself, after all, has had to re-engineer its business model since the traumatic (and predicted) flop of the IPO. Users flocked away to mobile, and Facebook had to redesign its primary user experience in order to cope. At the same time, the advertisers who are Facebook’s real customers lost interest in the brand-building display pages that were its key advertising product.

Facebook chased the audience, developing new advertising products to fit into the context of a mobile user experience. It worked: post-IPO Facebook has succeeded in getting revenue from its mobile advertising, so much so that it made $523m in net profits on revenue of $2.6bn in Q4. But it’s worth remembering that this represents Facebook concentrating on one very specific niche business: mobile apps discovery.

In the last quarter, 53 per cent of Facebook’s revenue, over $1bn, came from adverts for mobile app downloads. This is because you can expect to pay $2 in advertising costs for each install.

Facebook revenue by source June 2011 - December 2013
Wow. Note especially that the mobile apps line of business starts immediately after the IPO in June, 2012.
But this implies that revenue equivalent to something like a quarter of app sales through the iTunes App Store is going to Facebook just for ads. How long will Apple, whose train set it is, or Google, fundamentally an advertising business, put up with that before they launch something that competes?

The problem is summed up quite simply here:

“This exposes the strategic fallacy behind Facebook, which was the idea that there was going to be a monopoly on the social graph, and that Facebook was going to own it,” said Keith Rabois, a partner at venture capital firm Khosla Ventures. “That’s not true, and I don’t believe Facebook will constantly be able to buy its way out of this structural challenge.”

This pattern has been visible for a while. Rather than big multi-functional platforms, suddenly it seemed that leaner, focused, task-specific apps were in demand. We got Instagram, and Snapchat, and ask.fm, and Vine. It should come as no surprise that Unix-based iOS and Linux-based Android both seem to encourage app developers to do “one thing well” in the classic tradition of the core Unix/Linux utilities, with the unifying platform being the OS itself. So Facebook chased its audience again, buying Instagram.

Meanwhile, users sought out a new generation of mobile instant-messaging apps, which saw astonishingly fast growth and shocked the carrier industry with the hit to their SMS revenues. And Facebook has chased the audience again, with the WhatsApp acquisition.

WhatsApp: Doing One Thing Well

Whatsapp is nothing if not a lean, focused, task-specific app that does one thing well. Its USP could be summarised as “instant messaging done right”. Its business model is simplicity itself, asking users for a dollar a year, rather than seeking advertisers or volume-billing. The simplicity, as with most simplicity, is founded on engineering excellence – Whatsapp holds the record for the most concurrent TCP sockets, 2 million, on a FreeBSD Unix server. It’s because they spent the time and money developing their highly customised fork of the open-source ejabberd XMPP server that they kept costs down to the level where their business model made sense.

Across Europe, Whatsapp and the proliferation of other IM apps dragged SMS pricing down until it became a bundled, unlimited service. Vodafone, for example, bundles unlimited messaging with 1GB of data at its €29 price point. But, as we point out in the Future Value of Voice and Messaging strategy report, price is only part of the story. Price drove the acquisition of customers, but quality retained them. WhatsApp offers a searchable, conversation-based chat history and a one-tap voice messaging function; it remains shocking to this day that it took Apple to show SMS messages as threaded conversations like e-mail.

This Ars Technica interview provides some insight into the WhatsApp founders, their motivations, and their thinking:

“People need to differentiate us from companies like Yahoo! and Facebook that collect your data and have it sitting on their servers. We want to know as little about our users as possible. We don’t know your name, your gender… We designed our system to be as anonymous as possible. We’re not advertisement-driven so we don’t need personal databases.” This is more than a business position for Koum. “I grew up in a society where everything you did was eavesdropped on, recorded, snitched on,” he says.
“I had friends when we were kids getting into trouble for telling anecdotes about Communist leaders. I remember hearing stories from my parents of dissidents like Andrei Sakharov, sentenced to exile because of his political views, like Solzhenitsyn, even local dissidents who got fed up with the constant bullshit. Nobody should have the right to eavesdrop, or you become a totalitarian state–the kind of state I escaped as a kid to come to this country where you have democracy and freedom of speech. Our goal is to protect it. We have encryption between our client and our server. We don’t save any messages on our servers, we don’t store your chat history. They’re all on your phone.”

We see here a profound difference between Facebook and Whatsapp. Both of them let you share stuff with your friends, but the definitions of “share” and “friends” are radically different. WhatsApp CEO Jan Koum also seems to object to ads on principle.

Two Fundamentally Different Social Models

For Facebook, each post links a user with something (a photo, a link, an app, a product, a song) and with one or more recipients. You can organise these relationships as a directed-acyclic graph and use it to predict which users might be interested in which somethings. Advertisers will pay for this. As a result, Facebook is in the business of maximising the number of things you share and the number of people you share them with. You might never have considered telling some second-tier relative or distant acquaintance about this one weird trick, but Facebook is there to encourage you to do so, and indeed to suggest both people and things you might share. Oversharing, in a word, is its product.

For WhatsApp, this is basically beside the point. The app exists to let you contact the specific people you want to talk to, when you want to. It also lets you retrieve what was said later. In exchange for this service, it takes a small fee. The USP here is control.

These are two fundamentally different modes of social interaction. Facebook’s is all about surfacing as many of the weak ties between relatively distant people and things as possible; WhatsApp’s is about maintaining the strong ties between people who communicate all the time. Facebook is keen to encourage the casual, or even automated, sharing of pretty much anything – a “like” on a picture of someone’s cat – and therefore isn’t interested in how you might find it again. The infinite-scroll design of the front page makes this abundantly clear. At one point, Facebook even tried a feature that shared content without any user input whatsoever.

WhatsApp, by contrast, is all about conversations that you want to keep, that you entered into deliberately. It is surprisingly common for people to keep old mobile devices around in order to archive SMS, another example of how awful the clients built into them by vendors and carriers are. In Marshall McLuhan’s terminology, Facebook is a cool medium (like television) and WhatsApp is a hot medium (like telephony). Facebook requires half-attention, WhatsApp full attention.

So Where Are We Going?

One option would be full integration, with WhatsApp making advertisers its primary customer and probably going free to the end user, and the technology platform being integrated into Facebook’s. It seems hard, to say the least, for two such radically different products to integrate. Redistributing Facebook notifications and advertising into WhatsApp would be painfully spammy. Using WhatsApp messages or metadata in Facebook itself’s recommendation and ad-targeting machine would be a massive privacy intrusion in a political climate that has become intensely sensitive post-Snowden.

This is especially sensitive because WhatsApp uses phone numbers as identifiers; Facebook already tries to match user names to phone numbers whenever possible, something which has been quite controversial as it reveals the relatively intimate phone number to people you might not expect.

Users deleting their accounts en masse would be a very real possibility. WhatsApp’s use of phone numbers as identifiers makes a user exodus much more likely – users migrating to a different app wouldn’t need to find their friends list again, as would happen with an app that defines its own user IDs. And we know it can happen – Facebook’s acquisition of Instagram saw a sudden revival of interest in Flickr, the original and best photo-sharing site.

Facebook might consider running WhatsApp on in parallel, keeping it for the revenue. Some commentators have described this as trying to capture revenue from the telcos. We disagree. Experience shows that this is less about capturing market share than driving down prices market-wide, as SMS becomes a bundled service. That said, WhatsApp is a cheap business to run, so a large share of a radically reduced messaging revenue pool would be well worthwhile for it. This is less true of Facebook, though, as it is a substantially bigger company and one which is far more expensive to run.

Although $19bn seems very pricey from this perspective, it’s worth pointing out that Facebook stock is sufficiently in demand that its major shareholders, who after all include Mark Zuckerberg and most of the executives, won’t experience any perceptible dilution if the WhatsApp founders sell. Of course, Facebook doesn’t pay a dividend. Therefore, from their point of view, the $12bn in stock is un-money and the $3bn in deferred stock even more so. The $4bn in cash represents 18 months of Facebook’s net profits, which is considerably less scary.

This makes more sense than full integration, but it has its own problems. Other IM networks exist, and trying to raise prices in this ultra-deflationary environment seems futile. And what would become of WhatsApp as a Facebook division? The cautionary tale is surely Skype as a Microsoft company; although Skype kept its autonomy and its own CEO, it’s hard to think of any progress or innovation it has made since 2011. Instead, Microsoft’s own Lync has become deeply integrated right across Office 2013, Outlook, and Windows Phone.

Another way of looking at Facebook post-WhatsApp would be that the company is gradually moving towards something like Softbank’s business model (or Schroders in the .com boom), a conglomerate of Internet startups. In this model, Facebook shareholders would be buying exposure to the social startup industry more broadly, and Facebook as an organisation would primarily provide capital, data-centre engineering, and media sales functions to its acquisitions.

This might actually make a lot of sense. Facebook’s record of innovation in products, propositions, and design is rather chequered (an HTC First with Facebook Home, anyone?). However, its infrastructure engineering is on a par with Google’s or Amazon’s, and it has spent very heavily building a media sales force and engineering tools for advertisers. In many ways, the original Facebook product was an accidental innovation; why not concentrate on what the company does well?

A caveat, though, is that Facebook would probably struggle to resist the temptation to pour WhatsApp messages into the advertising machine, and even if it didn’t, the public might not believe its denials. Another caveat is that Facebook’s developer relations could be better; it has a bad reputation for making sweeping API changes that break other people’s businesses without consultation and also for competing with its partners. It has never really taken off as a development platform. And a further caveat would be that such a business model would be a lot like a VC fund, and investing in early-stage companies profitably is very difficult.

Ask the people who bought Facebook shares at the IPO, after all.