Late into the night on St. Patrick’s Day, Netflix hosted a makeshift “war room” at the company’s Silicon Valley headquarters. Engineers and product wranglers sat loosely around a long, U-shaped formation of desks, monitoring diagnostics and Hipchat conversations. Linguists and translators patrolled a ring of television sets to check subtitles and dubs in a dozen languages. Series star Charlie Cox was there, flanked by Netflix’s social media czars, firing tweets from his character’s account. They were all gathered for one of Netflix’s biggest moments so far this year: the season two premiere of Daredevil, which went live in 190 countries. At the same time.
This sounds like a major event. And in a way, it was. Just two months ago, Netflix debuted in 130 new countries—another simultaneous launch. The company that got its start delivering loaner DVDs in the mail is now positioning itself to be first truly global content network, and its original shows—especially marquee titles like Daredevil—are key to that ambition.
‘We know what everybody wants, which is the same catalog around the world.’ Netflix CEO Reed Hastings
But perhaps the even bigger deal is that sending a new show worldwide felt like no big deal at all; the Netflix war room was less Dr. Strangelove than late-night potluck. Champagne was poured, a countdown ended, the show began, and that was that. Absent the dim lighting and the bubbly and a gaggle of international journalists, it felt like just another day at the office.
That seeming ease of execution belies the time, money, and effort Netflix has sweated in the years leading up to this moment. From building up its global IT infrastructure to harnessing a world’s worth of data to creating content with global appeal, Netflix has been focusing its corporate energy on this exact moment.
For Netflix, becoming a truly global network presents a path for steady growth over the next decade and beyond. It has 75 million subscribers today; there are seven billion more out there. For the world, Netflix’s aspiration could mean much more: the first glimpse at what happens when every part of an online entertainment empire, from interface to content to delivery, is engineered to be everything to everyone, all of the time.
Why It Works
The moment Daredevil premiered, Netflix greeted its users with eight header image variations of Matt Murdoch and friends, shown to customers in eight identically sized chunks. Netflix immediately began tracking which top shots inspired the most streaming.
By now, those eight images will have given way to the best-performing two or three. After 35 days, one of those will become the default. The rest will vanish. This happens now for every Netflix original show. It’s survival of the clickest, all around the world.
Netflix is a notoriously data-driven company, and the Daredevil header art test is one of hundreds it will conduct this year. That data trove has also enabled Netflix’s gamble on global expansion,by illuminating one simple fact: People are all different, but not in the ways you’d imagine.
“There’s a mountain of data that we have at our disposal,” says Todd Yellin, Netflix’s VP of product innovation. Netflix has a well-earned reputation for using the information it gleans about its customers to drive everything from the look of the service to the shows in which it invests. “That mountain is composed of two things. Garbage is 99 percent of that mountain. Gold is one percent… . Geography, age, and gender? We put that in the garbage heap. Where you live is not that important.”
Stereotypes about what one region might like versus another are largely useless to Netflix. One might assume, for example, that Netflix’s anime streaming is heavily concentrated in Japan. Yet only 10 percent of the people who watch anime on Netflix live there. The other 90 percent, says Yellin, are distributed around the globe.
Think of your own block. Your neighbor binging Arrow doesn’t mean you’ve ever even heard of it. In fact, you may well hate it, if that same neighbor’s taste in blared 3 am music is any indication.
The upshot? Netflix uses a global algorithm. The titles you’re shown when you sign onto Netflix are just as likely to be influenced by someone from Abu Dhabi as Des Moines.
Freedom from worrying about signals like geography, gender, and age allows Netflix to hone its recommendations more sharply, and against less obvious criteria that for competitive purposes it doesn’t divulge. It’s what lets Netflix group its titles into a couple of thousand “clusters” based not on where people live, but what they like.
Netflix assigns each subscriber three to five of these clusters, weighted by the degree to which each matches their taste. “When you have more than 75 million people around the world, you can get really specific about who’s your taste,” says Yellin.
It’s survival of the clickest, all around the world.
For all the thousands of titles in the company’s catalog, the average member only sees 40 to 50 options in a typical visit. Clusters, which can comprise anywhere from tens of thousands to millions of subscribers, are what help ensure that those members see the right 40 to 50. You also aren’t limited strictly to your cluster assignments, Maybe you’d love Peaky Blinders—the algorithm occasionally offers glimpses outside your silo.
“We used to be more naive. We used to overexploit individual signals,” says Yellin. “If you watched a romantic comedy, years ago we would have overexploited that. The whole top of your screen would be more romantic comedies. Not a lot of variety. And that gets you into a quick cul-de-sac of too much content around one area.”
Getting that algorithmic balance just right is important to Netflix on a day-to-day basis, but absolutely critical when entering a new market. If geography were weighted heavily, if it mattered as much as we might assume, Netflix would have had a much harder time knowing what to show a person in a region it had never been before. Instead, it can use other proprietary signals as its guide.
The same principles apply to the Netflix user interface. There are small differences—viewers in Arabic-speaking countries will get their text right-to-left—but otherwise, the Netflix you interact with is the same as the one that tens of millions of people around the world do.
“User interface and design is the same everywhere in the world,” says Yellin. And whatever differences do exist get smoothed out through extensive testing. As WIRED has previously reported, in 2015 alone Netflix ran 160 A/B tests, primarily on new customers, each comparing between two and 20 different experiences. Since most of Netflix’s new customers are international, and have been for some time, its UI already acts as a global common denominator.
How It Works
You might think that a major premiere like Daredevil would threaten to strain Netflix’s infrastructure. In fact, as several war room engineers attested that night, it hardly makes any impact at all.
For starters, there’s no real traffic spike; it turns out that the people who watch Daredevil the minute it begins streaming would have been watching Netflix anyway. Even if there were a major surge in demand, the files for Daredevil were already sitting on local servers around the world well before the debut.
To understand how that’s possible, it’s important to realize that Netflix can fit inside a box. Specifically, a specially engineered “appliance” that acts as a localized content delivery network, stashing movies and shows at strategic locations around the world, usually close to high-activity areas. Think of it as a cut-off man in baseball. It’s faster, and more reliable, to toss content to your home base from nearby than it would be from one far-off, centralized location.
These so-called Open Connect appliances serve a simple purpose: To keep Netflix from clogging up the Internet. In North America alone, Netflix is singlehandedly responsible for 37 percent of downstream Internet traffic during peak hours. The service as a whole streams 125 million hours of content every single day. Without relieving as many pressure points as possible, things could get ugly, fast.
The total capacity of the Internet’s country-to-country backbone is 35TB per second, says Ken Florance, Netflix’s VP of content delivery. “Our peak traffic is more than that … . Our scale is actually larger than the international capacity of the Internet.” Netflix doesn’t literally break the Internet because the vast majority of its traffic is delivered locally, via Open Connect, rather than across the transoceanic cables that connect the Internet between continents.
A recent example: When Netflix launched in Brazil, it delivered 100 percent of traffic to that market directly from Open Connect boxes in Dallas and Miami. House of Cards got there all the same, but more slowly and at greater expense. Now, Netflix has worked with local internet service providers to install Open Connect appliances at critical points throughout the country’s data infrastructure. Netflix loads up each box—during off-hours—with one or two copies of what nearby customers are most likely to watch, based on internal popularity prediction models. That means one stream over a long distance, rather than thousands, if not more. If Netflix guesses wrong, and a substantial audience wants to watch something that’s not pre-loaded onto an Open Connect appliance, it takes Netflix just 15 minutes to move a show from Amazon’s cloud to one of its boxes.
The result? Eighty-five percent of Brazilian Netflix traffic today is local. In terms of practicality, it’s the difference between requiring a long-haul trucker and an Instacart courier. Around 1,000 of these boxes are out in the wild today, ranging from hulks that can hold the entire catalog to svelter units for more remote locations.
You’d think, given that scale, that bursting Kool-Aid man style into 130 new markets would require a significant number of new Open Connect locations. It took four: Dubai, Singapore, Hong Kong, and Johannesburg. That it took relatively little infrastructure to support such a broad expansion reinforces that presence doesn’t magically translate into success. If and when it gets significantly more global subscribers, Netflix can scale up, but instant supply doesn’t equal instant demand. America’s been sold on Netflix for years. Burundi, not so much.
Think Global, Act Local
In many ways, Daredevil is the quintessential Netflix property. It exhibits enough violence and language to give it a sense of edge—you won’t find too many bad guys hanging from meat hooks on CBS—while still falling comfortably short of triggering major parental concern. It balances critical success with pulp appeal. Most of all, though, it’s a Marvel product, which means it has built-in international recognition.
For today’s Netflix, that’s what matters most. You can’t go truly global, after all, unless you have content that can break out of wherever it happened to be filmed. That’s why Netflix handed Adam Sandler a four-picture deal. Ridiculous Six is a ridiculous film, but Sandler’s global popularity (and yes, he’s big in France), made it the most-watched movie in Netflix history within 30 days of its release. Or think of Narcos, a predominantly Spanish-language show from a Brazilian filmmaker, that found a sizable US audience.
But not every show is a slam dunk. Acquiring users isn’t a new challenge, nor is it limited to the developing world. Every region, even ones in which Netflix is well-established, presents uniquely shaped potholes to slow the company’s progress.
Take Canada, in which Netflix has operated longer than any country besides the United States. Canadians, says Netflix CEO Reed Hastings, harbor constant insecurities about the depth of the streaming catalog in their country, despite having access to several titles—Star Wars: The Force Awakens soon among them—unavailable to its southerly neighbors. In France, meanwhile, laws designed to protect theaters prevent the streaming of any movie within three years (yes, years) of its theatrical release. As such, it will be the only country denied access to The Big Short when Netflix starts streaming it this summer. And that’s just two countries.
“It may not sound like much, but multiply each unique exception across 200 markets, and it quickly becomes a major challenge,” says Brett Sappington, director of research at Parks Associates.
Taste profiles may be universal, but licensing agreements and legal frameworks are rarely so. And that’s before you even get to the most common complaint whenever Netflix enters a region: There’s not much to watch.
“We see this every time we launch a country,” says Hastings. “People want all the content in the world, and upon launch we don’t have X or Y. That’s a natural reaction.”
Netflix has taken steps to improve this situation, but not before making it significantly worse for a large number of users. Soon after the company’s January land grab, it announced that it would begin cracking down on the use of virtual private networks (VPNs) to access its content.
VPNs are good for lots of things, but for Netflix users they serve one primary purpose: To let you pretend you’re in a region you’re not, and access content that’s not locally available. This was especially helpful a few months ago, when it let people in over a hundred countries with no Netflix at all access the service.
The VPN squeeze comes now that those same people can access Netflix legitimately, though often with significantly smaller libraries than they’d had before. Not surprisingly, this has created a serious backlash, though it’s not clear if it’s had any material impact on actual subscriber numbers. Netflix doesn’t break those out by region and hasn’t updated its international and US figures since January.
“I think this may very well result in a couple of thousand subscribers shifting their money onto local services, but I don’t think VPN users was ever a major issue for Netflix,” says Ovum analyst Tony Gunnarsson. “I suspect the number of VPN users has been massively overblown in the media.”
Hastings, meanwhile, sees the VPN enforcement as an inevitable byproduct of going global. “Think of it as the maturation of all of Internet TV, as respecting those boundaries. We have the obligation to respect the rights that we buy,” he says. “It’s a simple fairness thing. If someone else is paying for the rights in Germany, we should respect it, just as we would want it in return.”
Since removing access to someone’s VPN makes their libraries smaller, the way to win those customers back is fairly straightforward: Offer more content, in more regions, to begin with.
That’s harder to accomplish than it sounds. Netflix has to compete with companies all over the world for the right to stream content. Those licenses can cover individual countries, or regions, or the entire globe, and can vary in duration. That’s why the Netflix catalog varies from place to place; it’s not uncommon for Netflix to win a bid a show in one country, but lose in another.
The more Netflix ups its bids worldwide, the more uniform its offerings will be from region to region. But these licenses can come at incredible cost. At the very highest end, Hulu paid $160 million for exclusive rights to stream all 180 episodes of Seinfeld. The deal will expire after five years, at which point a bidding war still start all over again.
That’s partly why in recent years Netflix has pared down its library considerably. Its US catalog has dropped by more than 30 percent since January 2014, according to AllFlicks. That’s at least 2,500 fewer titles in just over two years.
Despite that drop-off, Hastings says he remains committed to buying rights to third-party content. Still, the company’s focus has clearly shifted from filling in a jigsaw puzzle of licenses to focusing on those movies and shows it can stream exclusively on a worldwide basis. For the most part, that means putting money that used to go toward, say, showing a Sony-produced TV show in Europe, toward producing its own original content. It’s much easier to stream Daredevil to 190 countries at once if you’re the one who made it.
“We know what everybody wants, which is the same catalog around the world. We’re frustrated that we’re not there, but we will get there before everybody else,” Hastings says.
The next phase? Working with local content providers to make products with global potential. When Netflix does Bollywood, for instance, it will do whatever version of Bollywood it thinks has the best chances for success not just in India, but in Arizona.
That’s not always going to be easy. No Bollywood feature, for instance, has yet found mainstream success in the United States, and it’s optimistic to assume that all regional genres need to break out is the distributive powers of the Internet. The aspiration to universally appealing content also raises something of an existential question: If Netflix engineers a Bollywood production to the point that it has universal appeal, will whatever that looks like simply be a regression to the mean? Is Bollywood via Arizona still Bollywood at all? Or does making a movie or show that’s all things to all people just become mush?
It’ll be a long time before we know either way. While Netflix has actively licensed third-party content with local appeal, so far its non-US originals mostly comprise the Mexican telenovela Club de Cuervos and, arguably, Narcos.
And even if whatever does transcend borders ends up watered down, that’s not Netflix’s concern. The company gives its creative partners substantial leeway, and invests in crtitically lauded fare—think BoJack Horseman, or a recent multi-film deal with indie favorites Jay and Mark Duplass—but like any studio, it’s a business first. Netflix will spend $5 billion on content this year, allocated not based on artistic importance but potential subscribers gained. House of Cards wins Emmys, but Ridiculous Six sets records.
“It will take months, if not years, for these efforts to significantly contribute to the Netflix library and to acquisition of subscribers,” says Sappington. And that’s assuming a slate that’s more full of Orange Is the New Black hits—it’s a true international success—than Hemlock Grove duds. Then again, data-driven global recommendations and a global interface have generated success so far; why not global content?
The Long Game
Then there are all the things Netflix can’t control. Aside from Canadian self-doubt and French…ness, it faces blockades in many of the regions in which it recently launched.
Consider that in many emerging markets, Netflix is only available in English, priced in dollars, and only accepts international credit cards, or PayPal where available. That’s not especially appealing, particularly in areas where broadband penetration is already relatively low.
Then there are local governments, ISPs, and content to deal with. A protectionist streak in the former (hello, China!) can keep Netflix out of a country entirely, while a lack of interest from the former in Open Connect can result in a stuttering streaming experience. Brazil’s Globo, one of the largest commercial television networks in the world, and the largest in Latin American, has yet to license any content to Netflix at all. It’s been nearly five years.
Most of all, there’s competition. Ovum’s Gunnarsson estimates that in nearly every new Netflix country, the company will be greeted by up to 10 established video-on-demand services, many of them free, along with a handful of streaming on-demand services that look a whole lot like Netflix. These are companies that not only know what their market wants: they literally speak the language.
“Local market OTT services that have access to interesting local content have an advantage over Netflix,” says Parks Associates’ Sappington. In most cases, they also have a head start.
All of this is why, while Netflix added 130 countries this quarter, you shouldn’t expect its subscriber numbers to skyrocket in kind. “If Netflix proves successful in all of its 130 new markets—and going by past performance, it is safe to say it will—it will still take the company years to build the level of subscriber numbers it has in Western Europe and Latin America,” says Gunnarsson.
He expects the total member count to increase by up to 10 million this year; a large number, but not game-changing, especially the investment it has taken to get to the point where this kind of worldwide ambition was possible.
Still, if you’re Netflix, that’s probably fine. It took 10 years for it to stream its first video after a decade of DVDs, and nearly 10 more to build the scaffolding of a global streaming network. This is a company that’s not afraid of the long view.
“I think the goal is that we’re as strong in Vietnam and Thailand and everywhere else as we are in the US, in terms of awareness and usage,” says Hastings. “That’s 10 or 15 years away.”
Meanwhile, at the Daredevil war room, a shuttle arrived thirty minutes after the midnight launch, with another scheduled for the top of the hour. Most of the outside observers took the early option. Everything was running smoothly. There was nothing left to see.