“What do you think about gas in the tank for the long term?� asks Cindy Holland, Netflix’s vice-president of original content. It’s a Tuesday morning in May, and Holland and a handful of her direct reports are meeting in the 14th-floor San Junipero conference room of the company’s Hollywood headquarters. They’ve come to discuss renewal decisions for two existing shows, the Drew Barrymore–Timothy Olyphant zombie comedy, Santa Clarita Diet, and the recently launched remake of Lost in Space.

As Holland goes around the room, she stares at a laptop screen filled with the memos her team has prepared. She notes the mixed reviews for Lost in Space. “Do we care?� Not that much, it turns out. The show is renewed for a second season.

As they discuss story lines and other creative matters, there’s talk about “completion,� i.e., how quickly subscribers are moving through episodes to the end of the season. Holland quizzes the room about how the shows are doing internationally and if they’re under- or overperforming in certain territories. Someone mentions that Barrymore and Olyphant traveled to the Philippines to promote season two of Santa Clarita: “It’s the first time we took a show there,� she says, adding that the promotional support seemed to pay off: “We’re really, really excited about the fact that it’s traveled globally.� There’s enough gas in the tank, they decide, for a season three.

The conversation moves on to new projects, including Away, an unannounced drama from creator Andrew Hinderaker (Penny Dreadful) and executive producers Jason Katims (Friday Night Lights) and Matt Reeves (Cloverfield) that revolves around an international group of astronauts on the first-ever mission to Mars. “Do you have a clear sense of who is that core fan base?� Holland asks. “I feel it’s a pretty global show in terms of the cast and the diversity of players,� says one executive. “But I also think because there’s that epic love story at the center, it’s going to attract a female audience.� “You probably also get the sci-fi audience as well, right?� Holland says. “I don’t think we’re going to get a hard-core sci-fi action audience,� the executive replies. “That’s not what this is.�

Also on the agenda is a not-yet-announced limited series. There’s a brief debate over which of Netflix’s many content “verticals� it will fall under. “It’s kind of a hybrid between series and film in terms of the biopic nature,� one executive says. “Right now, it’s projected somewhere between period romance and the black-film vertical,� says another. Adds someone else, “It doesn’t fit squarely in either, so we think there’s a nice in-between.�

The meeting ends in less than an hour, and the futures of four of the roughly 1,000 original titles Netflix plans to make (or acquire and distribute) this year are a bit more certain.

Netflix’s overthrow of television’s old business model began just seven years ago. That’s when the Silicon Valley company best known for mailing DVDs in little red envelopes outbid AMC and HBO for the rights to a drama from director David Fincher, a remake of the British mini-series House of Cards. It was a big deal at the time, both because of the money Netflix was spending ($100 million for two seasons) and because it was the first hint of the streaming platform’s ambitions to evolve beyond a digital warehouse for other conglomerates’ intellectual property.

House of Cards is airing its final season this fall, and Netflix now makes more television than any network in history. It plans to spend $8 billion on content this year. “I’ve never seen any one company drive the entire business in the way Netflix has right now,� says Chris Silbermann, managing director of ICM Partners and agent for Grey’s Anatomy and Scandal creator Shonda Rhimes, who moved her production company to Netflix last year.

TV has gone through major transformations in the past — cable and Rupert Murdoch’s Fox toppled the hegemony of the Big Three broadcast networks in the 1980s, for instance — but this leap dwarfs all others. Netflix doesn’t want to be a streaming, supersized clone of HBO or FX or NBC. It’s trying to change the way we watch television. Whether it can do that while turning a profit is another matter, given the more than $6 billion in debt it’s amassed during its expansion. But Wall Street seems optimistic: In recent weeks, its overall market capitalization has at times grown past $150 billion, surpassing Disney to become the most-valued media company in the world.

CEO Reed Hastings and tech entrepreneur Marc Randolph launched Netflix in 1997, rolling out its DVD-by-mail service the next year and introducing the all-you-can-watch subscription model in 1999. The service has offered streaming since 2007. But it was the company’s move into original content that has upended so many norms of the TV business: Netflix doesn’t waste millions making pilot episodes of shows that will never air; instead, almost every project it buys is purchased with the intention of going straight to series. It invented the idea of binge-releasing — dropping full seasons of shows all at once, rather than doling out episodes week-to-week, as TV had done since I Love Lucy. Instead of selling its content to international partners, Netflix has eliminated global middlemen and set up shop in over 190 countries, allowing it to debut a new season of an American animated series (BoJack Horseman) or a German thriller (Dark) around the planet, all on the same day and at the same time. It has replaced demographics with what it calls “taste clusters,� predicating programming decisions on immense amounts of data about true viewing habits, not estimated ones. It has discovered ways to bundle enough niche viewers to make good business out of fare that used to play only to tiny markets.

And shareholders have given it the money to poach the top showrunners from ABC (Rhimes) and FX/Fox (Ryan Murphy), committing upwards of $400 million to deny those networks their biggest hitmakers. It’s greenlit series from the past two Oscar-winning directors (Damien Chazelle, Guillermo del Toro) and today’s most successful producer of network sitcoms (Chuck Lorre, whose next show for the service stars Michael Douglas and Alan Arkin). Netflix has also handed out paychecks worth, in some cases, more than $20 million to a constellation of stand-up stars (Chris Rock, Dave Chappelle, Ellen DeGeneres), signed the next generation of talk-show hosts (Michelle Wolf, Hasan Minhaj), and given a new home to older ones (David Letterman, Norm Macdonald). And last month, it announced a deal with Barack and Michelle Obama to make TV shows and movies.

“The first word out of everybody’s mouths in meetings is, ‘How do we deal with Netflix?’ � says one longtime TV-industry executive. “‘How do we compete with Netflix? What are they doing?’ � Disney’s pending purchase of much of 20th Century Fox’s film and TV assets — which has prompted a counterbid by Comcast, parent company of NBCUniversal — is in no small part a reaction to the rise of Netflix. Robert Iger, Disney’s CEO, wants the added scale 20th Century Fox’s assets will bring as he prepares to launch Disney’s own direct-to-consumer streaming service next year. The proposed AT&T–Time Warner merger is similarly designed to help AT&T take on Netflix.

Mysterious though it may seem, Netflix operates by a simple logic, long understood by such tech behemoths as Facebook and Amazon: Growth begets more growth begets more growth. When Netflix adds more content, it lures new subscribers and gets existing ones to watch more hours of Netflix. As they spend more time watching, the company can collect more data on their viewing habits, allowing it to refine its bets about future programming. “More shows, more watching; more watching, more subs; more subs, more revenue; more revenue, more content,� explains Ted Sarandos, Netflix’s chief content officer. So far, it’s worked spectacularly well: Netflix has gone from around 33 million global subscribers before House of Cards premiered to over 125 million today. Wall Street analysts have predicted Netflix could flirt with 200 million subscribers by the end of 2020; by 2028, one Morgan Stanley analyst has said, 300 million is possible. “The thing that keeps me up at night is scale,� says Sarandos. “It’s a mind-boggling amount of programming that’s being produced here. How do we keep scaling it?�

One answer is cultural. “I’m building a team that’s oriented as saying ‘Yes’ in a town that’s built to say ‘No,’ � Sarandos says. That’s not just New Age–speak. It’s practical. To stimulate volume, Sarandos and Holland have put in place an extraordinarily decentralized development and production pipeline, one that allows Netflix to operate like ten or 15 semi-independent entertainment companies — whose output all happens to be distributed by a single service.

“Two layers beneath Cindy have full greenlight� authority, Sarandos says. “The only way that we can do what we do at the quality and volume we’re doing it is to give power to my executives to make those choices.� One agent I spoke to told me that translates to at least “five or six� scripted-development executives he can pitch knowing they have the authority to make a project a reality. The heads of Netflix’s other big divisions — international, unscripted, documentary, stand-up comedy — are similarly able to give an idea the go-ahead. “Most of my team have more buying power than anyone has selling power in Hollywood. My direct-report team can greenlight any project without my approval. They can greenlight it against my approval!� says Sarandos.

I ask Sarandos to give me an example of something that’s gotten made over his objections. He cites What Happened, Miss Simone?, the documentary from director Liz Garbus. Lisa Nishimura, Netflix’s VP of original documentaries and comedy, was a big proponent of the film, but Sarandos wasn’t convinced. “We fought about it for six months,� he recalls. “She came in once or twice a week to say why she had to make this movie, and I would tell her that it’s too expensive, and music docs don’t play very big. She’d come back and explain to me why this isn’t a music doc. She was 100 percent right and I was 100 percent wrong. That was an incredible movie, and as soon as it started delivering, I felt like it was a big miss for me to have held it back that long.� Sarandos was similarly iffy on American Vandal, last summer’s comic mockumentary that ended up being a word-of-mouth hit. He kept telling the development team he didn’t think it made sense; they made it anyway, and now Netflix is working on a sequel.

Lower-level executives aren’t completely free agents. “They have some budget constraints,� Sarandos says. “Somebody who typically can greenlight a $3 million show, but has a $10 million show [under consideration] — they’re going to check first. Cindy will bring things to me that seem [riskier] and be like, ‘Hey, this is why we’re going a bit further on the limb with this one.’ �

“This [idea] that if you have volume, you can’t have quality?� says Holland. “I think it’s convenient for people who are limited by time slots or budget. If you can have one network that has a dozen shows and they’re good quality, why can’t you have the equivalent of four networks with a dozen shows each? Why can’t you have more than that? We have the ability to support a larger number of artists than most people can.�

While there’s still room to grow domestically, Netflix’s biggest opportunity for scale is overseas. On a Monday morning in April, I attend a meeting run by Erik Barmack, head of the company’s international-originals team, where a couple of wall-mounted video screens show the names of various staff phoning in, as well as a live feed from the Amsterdam office. Netflix has a division devoted to acquiring foreign programs from networks like the BBC, but Barmack oversees the production of original non-English-language shows made for Netflix outside the U.S., including Dark (Germany), Ingobernable (Mexico), and 3% (Brazil). A number of Netflix American-made originals are popular outside the States — “As a percentage of total watchers, as many people watch 13 Reasons Why in India as watch it in the U.S.,� Sarandos tells me — but in order to compete and grow in foreign markets, Netflix believes it needs to offer subscribers stuff made in their own countries, by local artists.

Netflix’s international push is grounded in lessons the company has learned from its expansions into genres once thought to have limited appeal to American audiences. Big numbers in niche categories prompted Lisa Nishimura, VP of original documentaries and comedy, to suggest the company start making content in those areas. “We started to ask, ‘Is it really niche, or have the distribution channels for those categories been historically disaggregated, making it difficult to actually get scale and momentum and word-of-mouth and all those things that help to grow audiences over the course of time?’ � Nishimura says. “On the documentary side, people pointed to box office to say, ‘See? It’s tiny.’ What I contended was, that was a reflection of how many people watch a documentary on Friday night in that particular moment in time, not the potential of the actual audience size for the documentary if you made it easily available.�

Back at his meeting, Barmack is talking about the success of one of his international productions, the Spanish crime thriller La Casa de Papel (known in the States as Money Heist). He lets the group in on some discussions he’s been having about the show with Sarandos. “Ted and I were looking at the peculiarity of Casa — how huge it is everywhere that is not English-speaking and then relatively small in the U.S. and U.K. and Canada,� he says. It’s not that Netflix isn’t happy with the show’s American performance: Sarandos actually touted its audience levels here during an April earnings call with investors. But Barmack says he and Sarandos have been wondering whether there’s a way to do versions of the show in the U.S. or U.K. — something Netflix has heretofore avoided because English-language adaptations of international fare often lack the charm of the originals or end up being repetitive, particularly since American audiences would’ve had access to the foreign versions. “What we’re debating is not whether we should just take the script line by line of Casa and put it in the U.S.,� Barmack explains. “But is there something unique we’ve tapped into, something around the heist, around the characters in the show having city names, the humor, and whether we can make a version of it in the U.S. and the U.K.?�

Barmack’s staff, many of whom hail from outside the U.S., don’t seem keen on the idea. “I think this group has been doing such a great job making non-English stories relevant that by turning these non-English stories into English we may be shooting ourselves in the foot,� one development executive tells Barmack. “We’re making a point that there can be great fiction that is not English. When we turn it into English, we’re giving it the status of being not as important.�

Later, I ask Barmack how he would move forward given the pushback. “For me, there are three buckets,� he says. “The data, the art, and the regional sensitivity. I’ll look at the data, and if everyone’s against it, it’s unlikely I’ll do something. If it’s a 50-50 thing, I’ll just make the call.� While some were opposed to repurposing the themes of Casa in English-language shows, Barmack tells me that after I left the meeting, one of his Mexican staffers, a former producer named Francisco Ramos, mentioned an idea Casa co-creator �lex Pina had pitched. “He said, ‘Look, �lex’s vision is that with every season, more and more will be at stake. We actually may want to take the show to the U.S., and that same crew would steal from Fort Knox,’ � Barmack says. “And I was like, ‘Boom.’ That’s both sensitive, we’re in character, maybe we call it a spinoff, and it’s loyal to the vision of the story. That would be amazing.�

One of Hollywood’s great unsolved mysteries of late has been how Netflix determines whether a show is a fit with its brand. Most traditional networks, even those aimed at a relatively broad swath of viewers, use their programming to define carefully crafted identities. HBO, FX, and AMC, for example, want critical acclaim and Emmys, even if that sometimes means a smaller audience. Sarandos, however, says he doesn’t like to limit his options. “I don’t want any of our shows to define our brand, and I don’t want our brand to define any shows,� he tells me, sitting in his office, where not one but three posters for The Godfather hang. “There’s no such thing as a ‘Netflix show.’ That as a mind-set gets people narrowed. Our brand is personalization.�

That quest for personalization explains why Netflix’s U.S. originals keep nosing into new genres. A couple years ago, some of Holland’s team saw an opportunity to reinvent young-adult dramas; out of that push came last year’s hit 13 Reasons Why and this year’s buzzy coming-of-age comedy On My Block. The company is also ramping up its unscripted offerings after realizing how popular the form was with its users. “One in three subscribers watch Netflix unscripted shows monthly,� says Netflix content VP Bela Bajaria, whose team scored an early victory with its Queer Eye reboot.

Because of Netflix’s Silicon Valley roots, many assume the company’s vaunted algorithm is where its decision-making process starts. But at almost every opportunity, Sarandos and Holland downplay the role data plays. “You have to be very cautious not to get caught in the math, because you’ll end up making the same thing over and over again,� Sarandos says. “And the data just tells you what happened in the past. It doesn’t tell you anything that will happen in the future.�

One case where data proved wrong: The End of the F***ing World, the British-made dramedy written by Charlie Covell, which was released internationally just after New Year’s Day. “It was astounding how popular it was for us,� Sarandos says, explaining that the company’s internal forecasts fell well short of how the show actually performed, not just in the States but around the world. “On one level, it was a massive failure that we didn’t see that coming,� he says. Sarandos asked his team, “What else are you highly confident about that you can’t see coming?� (At least they won’t underestimate Covell again. They’ve just given her a ten-episode order for Kaos, a dark hourlong comedy reimagining Greek mythology.)

So far, the company seems to have resisted the temptation to use what it knows about subscribers to guide the creative direction of its shows. Eric Newman, one of Netflix’s more tenured collaborators — he produced Hemlock Grove before becoming showrunner for Narcos and a producer on Will Smith’s feature Bright — says he sometimes wishes Netflix execs would give more data-driven feedback. “I ask for it and never get it,� he says, laughing. “I am always like, ‘Who do people like? You guys know everything! You know when someone rewatches a segment of a show, where people turn things on or turn them off.’ It would be really interesting for me to know.�

And yet, Big Data is unquestionably part of the DNA at Netflix. Holland admits that even when a big star or producer walks into the building with an undeniable pitch, she and her team are likely to take a beat because they want to do their homework — including looking at data — before giving a thumbs up. In some ways, how Netflix uses its internal subscriber data isn’t much different from the way a traditional TV network leans on Nielsen ratings. When This Is Us premiered on NBC and drew a big audience, executives there (and at its competitors) immediately put out a call for more tear-jerking family dramas. But Netflix’s data allows it to be vastly more precise, giving it an enormous competitive advantage. “We have projection models that help us understand, for a given idea or area, how large we think an audience size might be, given certain attributes about it,� Holland tells me. “We have a construct for genres that basically gives us areas where we have a bunch of programs and others that are areas of opportunity.�

Netflix calls these groupings of similar programs “verticals� — super-specific genres of film and television, such as young-adult comedies, period romances, or sci-fi adventures. Traditional networks that aim for broad audiences obviously try to fill their schedules with shows from multiple categories, too: HBO’s lineup features everything from satirical comedies (Veep) to sci-fi thrillers (Westworld). What’s different at Netflix is the company’s desire to fill so many of these categories with content. When gauging a show’s performance, Netflix will consider how big its audience is and whether the show is cost-effective. But it also cares about whether a show is performing well across multiple verticals, since that means the series is reaching a larger number of “constituent groups,� as Holland calls them during one meeting.

If verticals are the way Netflix executives think about what kinds of content to buy or make, taste clusters help them analyze how subscribers interact with programming. The phrase, along with the interchangeable “taste communities,� comes up time and again during my visits. Instead of grouping members by age or race or even what country they live in, Netflix has tracked viewing habits and identified almost 2,000 microclusters that each Netflix user falls into. While it’s not a direct parallel, taste communities are sort of like Netflix’s version of the demographic ratings used by traditional ad-supported networks, just more evolved. Because their business model is heavily weighted toward pleasing advertisers, broadcast and cable outlets such as NBC or Lifetime rely on demos — women under age 35, men 25 to 54, African-Americans 18 to 49 — to make sure their series are resonating with groups of viewers coveted by advertisers.

Netflix also wants programming that appeals to distinct groups of people, so in theory it could have a use for demos, too — and early on, it did. Netflix VP of product Todd Yellin tells me that when he got to the company about a dozen years ago, he figured demos would help him create a more personalized experience for subscribers. “Nielsen gets it, the networks get it, I should know the age and gender,� he says he thought to himself. “I asked people signing up, a long time ago when we were a DVD service, ‘What’s your age and gender?’ We would use that information to recommend shows.� Yellin found that age and gender were far less reliable in predicting future DVD requests than a user’s past viewing history. “Nowadays, in our modern world, hit play once and it tells us volumes more than knowing you’re a 31-year-old woman or a 72-year-old man or a 19-year-old guy,� Yellin says. Or, as Sarandos puts it, “It’s just as likely that a 75-year-old man in Denmark likes Riverdale as my teenage kids.�

Where taste communities and Nielsen demos differ is in the way they’re used. Demo ratings are how linear networks measure success; taste communities are the tool Netflix relies on to drive viewers to new material it estimates they might want to watch. The best explanation of how this works that I heard during my time at Netflix came during a new-showrunner orientation. Whenever new producers start a project, Netflix gathers a dozen or so representatives from its various departments to give them a presentation on how the company works. The Tuesday afternoon I sat in, the team from writer Susannah Grant and producer Katie Couric’s upcoming limited series Unbelievable was getting the rundown.

Olivia De Carlo, a director on Netflix’s originals product launch strategy team, explains to Grant and her partners how data — specifically taste communities — helps Netflix target shows to specific audiences. Once Unbelievable goes on the service, Netflix will use data to connect the show “with the people that are going to love it,� De Carlo says. “Sometimes I will liken this to a matchmaking service, and think about people dating their content and content dating people. We’re able to think about all the great information we have about how all those dates went, what ended up being a long-term relationship, what was a one-night stand — and we use that to think about how we set people up with their titles.�

The Netflix algorithm figures out which taste communities a member is in and then pushes the shows it thinks those members will enjoy to the top of their home screen. “We have a saying: Your Netflix is not my Netflix,� De Carlo says, noting that taste communities aren’t some static construct, either. “Most people are usually members of a few different communities,� she says. “We’re complex beings, we’re in different moods at different times.�

To show Grant’s team how this works, De Carlo gives a PowerPoint presentation. It shows how one of Netflix’s biggest hits, Black Mirror, plays particularly well in two major taste communities: Cluster 290 and Cluster 56. “We didn’t come out of the gate and say, ‘We think Black Mirror is for this audience or not for that audience,’ � she says. “But after we launched the show, we’re able to start to see patterns.� The chart shows how folks who liked Black Mirror were also fans of Lost and Groundhog Day. “On the surface, if you thought about Groundhog Day with Black Mirror, you might not find an obvious similarity,� De Carlo tells the group. “Lost and Black Mirror is also a stretch. But when you look at these in aggregate, you can see this through-line of supernatural or extreme worlds, and somehow that clustering tends to make more sense.� She then points to another graphic detailing other shows and movies that “have this dark-drama through-line,� and spells out what it means to the Netflix recommendation engine. “If a member hasn’t yet watched Black Mirror but they’ve watched Shameless and Orphan Black and The OA, we can be relatively confident in suggesting Black Mirror to them,� De Carlo says. Linear networks have their own way of driving audiences to shows. A good time slot or a promo during The Big Bang Theory can get viewers to sample something new. But linear TV’s promotional tricks are inefficient and miss out on wide swaths of potential viewers.

Another day, Yellin shows me a graphic featuring more than a dozen different images from the 1980s-set women’s-wrestling comedy GLOW. The company calls these mini-posters “row art� — they’re the little squares you see as you’re scrolling through Netflix and deciding what to watch. Yellin describes what’s on the screen: “Here’s Alison Brie.There’s Marc Maron. There’s a woman’s lips; there’s just two cats fighting,� he says. He points out the numbers underneath each image, explaining that they represent what percentage of Netflix’s 125 million subscribers are seeing that particular image when they come across GLOW on the app. “Eight percent for this one with Betty Gilpin; the Marc Maron one gets 6 percent,� he says. Whenever a new original premieres on Netflix, Yellin’s team will start off by randomly assigning different images to different subscribers, using those taste clusters as an initial guideline. “We try to find the images that unlock the content for people out in the world who have similar tastes to you,� he says. “Not the people who live in the next apartment or house over from you, not the people who live in the same Zip Code, not even the people who live in the same country: the people who tend to enjoy your kind of content. What image is unlocking them to play GLOW or Ozark?�

Despite its seemingly ever-increasing budget for originals, Netflix has been known to cut its losses by abandoning projects that don’t work. While Sarandos says 80 percent of the service’s shows get renewed, it’s no longer unheard of for it to cancel a series after just one or two seasons. Netflix is even starting to cull shows with good reviews, like Lady Dynamite, the Maria Bamford comedy, whose second season scored 100 percent on Rotten Tomatoes — and still got yanked. “I’m building a team to say ‘Yes,’ but we are also responsible shepherds of our members’ money,� Sarandos says. “Relative to the size of the watching of a show, we want the show buys to make sense.�

It might sound like Netflix wants all of its comedies to be as broad as Fuller House or its dramas to be four-quadrant hits like Stranger Things. But Sarandos says that’s too simplistic. “Nothing is too niche,� he says. “It’s just relative to what it costs. We can put a smaller show on the air and support the economics through subscriptions, but it’s not infinite. Eventually, there’s opportunity cost.� Smaller shows are more appealing to Netflix if they’re not super-expensive and they reach an audience not otherwise being served by more popular programming.

In the case of Lady Dynamite, while the show wasn’t prohibitively expensive, Sarandos suggests it simply wasn’t reaching enough of the viewers Netflix expected would be into it, even after two seasons. “Lady Dynamite, I would say, is a show I was as passionate about as anything on Netflix,� says Sarandos, and while it’s not uncommon at some TV networks for shows to get renewed because executives are fans, “we don’t do that,� he says. “We try not to program for ourselves. That’s the key. We’ve had to cancel shows that I’ve loved.� But given how many dozens of series and specials and movies Netflix now greenlights every year, couldn’t Netflix do, say, one or two fewer $20 million stand-up specials and make a season three of a Lady Dynamite? “Yeah, but our fans are trusting us to spend their subscription money well,� he says. “Those are the choices we have to make for them. And they can cancel Netflix with one click.�

Two shows that make for an interesting case study: One Day at a Time, a remake of the Norman Lear classic recently renewed for season three; and Everything Sucks!, a ’90s-set coming-of-age tale canceled less than two months after its first season launched. Both have modest budgets and lower-than-expected viewership. So why did one die a quick death while the other will live another year?

I ask Holland about Sucks, which one talent-agency insider tells me was produced by Netflix for under $1.5 million per episode — or less than $15 million for the entire first season. (“A season of that show cost like two-thirds of an episode of The Crown,� my source says.) Unfortunately, even at that low price point, Sucks simply didn’t generate enough interest from subscribers. “It sucks that it didn’t have a broader audience,� Holland says, leaning into the pun. “We couldn’t get out of that core appeal.� But it wasn’t just that not enough people watched. Among those who did, “a smaller-than-average number were completing it,� Holland says.

It turns out that for Netflix, having too many subscribers turn a show off midway through an episode and never return, or watch a couple episodes and then bail, might be just as bad, or maybe even worse, than not having a big audience, period. Sarandos later gives a name to this metric: survivorship, or “Did people who started watching episode one keep watching?� In the case of Sucks, “not a lot of people did.�

Netflix doesn’t necessarily care if you binge-watch an entire season of a show within a couple days of it launching. “We’re not trying to encourage that,� Sarandos says. “The completion of a single episode is a more important trigger. We wouldn’t be looking at, ‘Are people plowing through it in the first weekend?,’ because the number of people who do that is pretty slim.� But one metric I heard repeatedly during my visits to Netflix was 28-day viewership — basically how many people completed a full season of a show within the first four weeks it’s on the service. Sarandos also tells me the company looks at which shows new subscribers watch first: It lets them know if a show is driving people to sign up for Netflix.

Sarandos and Holland tell me, again, that while data is a tool for them, their various projection models and cost analyses don’t dictate their decisions. “It’s 70 percent gut and 30 percent data,� Sarandos says. “Most of it is informed hunches and intuition. Data either reinforces your worst notion or it just supports what you want to do, either way.� The company also sometimes ignores the data if executives have enough passion for a new project, Sarandos says, calling such cases “forward bets, where you go to a full season even though the model’s not quite there.�

That may be one reason One Day at a Time is still around. “It has a good audience,� Holland tells me. And yet, without getting specific, she seems to confirm a perception shared by the show’s production team and other TV-industry sources: Despite strong reviews, One Day has not become the breakout success Netflix had hoped. “We have been a little perplexed as to why its audience base hasn’t broadened beyond that core passionate fan base,� Holland says.

What seems to have made the difference for One Day is that the show has the right kind of audience for Netflix. I ask Sarandos what factors get discussed when debating the fate of a show such as One Day. “The unique value that it serves,� he says, is that “there were five or six different viewer groups who have a very strong affinity for that show. Latino for sure, LGBT for sure, female for sure. And it tells a different story than is being told on Netflix.� (By the way, since Netflix doesn’t do demos, he says he doesn’t know for certain that any of those groups are watching One Day, “but from social media and from other [metrics] you can kind of triangulate it.�)

Netflix’s approach to what it orders, renews, and cancels even applies to some of its top talent. Jenji Kohan (Orange Is the New Black, GLOW) signed an overall deal with the company last year, and yet Netflix has told her no, and more than once. “I’ve certainly pitched shows they’ve turned down,� she tells me. “I’m doing a lot of projects now with writers who’ve come up through my rooms and have brought them [to Netflix], and sometimes they don’t bite.� Last year, before becoming exclusive with Netflix, Kohan teamed up with actor-writer Jamie Denbo to pitch the company a comedy-drama called American Princess, about a woman who ditches the Upper East Side to join a Renaissance fair. Holland and her team turned down the idea.

“That’s showbiz,� Kohan says. “I think they thought it was too niche.� Does she think the project was killed because Netflix’s data told the company it didn’t make sense? “I know they love them their algorithms,� she says. “But I mean, they’re also people. So when we’re pitching, they have to connect with the material before going to the engineers and saying, ‘Who will this appeal to?’ �

I ask Holland why she and her team rejected American Princess. “We ultimately felt that Renaissance fairs — although there is an audience for them in the U.S. — it’s kind of an unknown, strange novelty outside of the U.S.,� she says. “So we felt that it probably wasn’t going to be a particularly global show for us. We also had a lot of questions about how you maintain legs on that story over many seasons.�

The good news for Kohan and Denbo is that American Princess is still getting made. When Lifetime heard the pitch, the network gave it a Netflix-like straight-to-series order, with production set to start this summer. Kohan doesn’t take any delight in Lifetime’s having said “Yes� when Netflix turned her and Denbo down; quite the opposite. “I wish it were at Netflix,� she says. “I really don’t like having to deal with a cable network at all. I’m probably biting the hand that feeds me for saying that.�

Sometimes outside factors weigh on a programming decision. That was the case with last year’s call to fire Kevin Spacey from House of Cards following allegations of sexual harassment and assault — and then keep the show around for another season without him. Sarandos tells me getting rid of Spacey wasn’t difficult. “The hard part was, what about the 300 people who make this show?� he says. “What about the 2,000 people in Baltimore who depend on that show for their living?� After Spacey’s firing, Sarandos and Holland reached out to the show’s producers and star Robin Wright to let them know Netflix was open to “a Kevin-free solution for the last season,� if they could find the right idea. “We liked what they came up with and we were able to move forward with it quickly,� Sarandos says. “It delayed the launch of the last season, but I think that the fans are going to be very happy.�

But he won’t say how many fans. During my visits, Netflix refused to budge from its policy against revealing exactly how many people watch specific shows. “Is Stranger Things the biggest show on Netflix?� I ask Sarandos during one meeting. He laughs. “It’s up there!� Later, I ask him how many potential viewers Netflix has, since most of its 125 million-plus paid subscriptions are obviously used by more than one person. “About 300 million,� he says. Given that, and the platform’s international reach, couldn’t one of Netflix’s shows eventually reach 40 or 50 million viewers? “Yeah, of course,� Sarandos offers. Has that already happened? “Definitely,� he says.

To answer my questions about the relative popularity of shows without actually answering them, Sarandos shows me a chart he’s printed out of the most popular TV shows as ranked by IMDb users. While the accuracy of the site’s ratings has been questioned in the past, Sarandos says IMDb is a “good indicator of what works on Netflix, because it’s a pretty net-savvy, entertainment-centric person that gives feedback. It’s better than Rotten Tomatoes.� The chart lists the top-30 new shows of the 2016–17 TV season. “Fourteen of them are Netflix original shows,� he brags. “Now this is global, so like Riverdale is a CW show [in the States], but it premieres as a Netflix original somewhere else. No one else on this list has more than three shows.�

Later, Sarandos cites IMDb again as evidence for the success of one of Netflix’s original movies, the teen-targeted romantic comedy The Kissing Booth. Sarandos calls it “one of the most-watched movies in the country, and maybe in the world� — but of course he won’t offer me any internal data to back that up. “In [IMDb’s] popularity rankings right now, it’s the No. 4 movie behind Deadpool 2, Avengers: Infinity War, and Solo,� he says. “Jacob Elordi is the male lead. Three weeks ago on the IMDb Star-o-Meter, which is how they rank their popularity, he was No. 25,000. Today he is the No. 1 star in the world. And Joey King, the female lead, went from like No. 17,000 to No. 6. This is a movie that I bet you’d never heard of until I just mentioned it to you.� Sarandos’s point: Because reporters like me don’t have ratings or box-office numbers, we’re too quick to listen to rivals who claim stuff on Netflix is getting lost. “This is the competitive message you hear out of a couple of different networks and studios all the time. It is so wrong,� he says.

Netflix won’t even tell its creators how many people watch their shows — which was actually a selling point for recent hires Ryan Murphy and Shonda Rhimes. “Ever since Glee, it’s been a daily death having to get up in the morning and get your daily report card that you know is a lie,� says Murphy. “The people who watch my shows are nontraditional TV viewers who are interested in nontraditional fare. They’re not going to watch something when you tell them to; they’re going to watch it when they want. I have so many tearful conversations with actors, having to say, ‘[The overnight ratings are] not the true story.’� Rhimes says she never paid attention to Nielsen numbers, even when they supposedly mattered. “What I like is that now I don’t have to work at a place where people believe [ratings] could be helpful for me in some way, send me those numbers, and expect me to translate them into anything,� she says.

Sunny as things seem when you’re inside Netflix, there are those in Hollywood who are not so enamored of the company and its appetite for showrunners, stars, and former leaders of the free world. Its competitors argue that Netflix’s volume and quest for scale don’t allow for quality control, that one company simply can’t support so many TV shows and movies without risking some of those projects’ vanishing into the void the moment they’re released. And there’s a nagging fear that the company’s promise to consumers of video nirvana is just part of a plan to push competitors out of business, the way Amazon and Walmart have put the squeeze on so many former rivals.

Sarandos has said he’ll be selective about doing more deals like the ones he cut with Rhimes and Murphy; Black-ish creator Kenya Barris could be the next, though it’s not yet a certainty. (As for the former First Family, “when we heard the Obamas were forming Higher Ground Productions, we wanted Netflix to be their home,� Sarandos says. “I didn’t want to see that deal go anywhere else.�) Such deals, of course, mean rivals have to pay more to keep coveted creators from defecting, leaving less for them to invest in other areas.

Still, Sarandos insists he’s not trying to put anyone out of business, arguing that Netflix doesn’t need others to fail for it to succeed. “We’ve gotten to the level that we have without much attrition on that side,� he says. “There’s been some cord-cutting, some cable-thinning, I guess they call it. But HBO hasn’t lost a single [subscription] since our growth has gone up. Part of our ability to keep growing is based on having a healthy competitive landscape.� Netflix also doesn’t win every project, Sarandos says. “We get outbid all the time — all the time,� he says. Indeed, one senior TV agent told me that Netflix aggressively pursued a Jennifer Aniston–Reese Witherspoon drama about a fictional morning-news show but lost out to Apple’s fledgling TV division.

Fears that Netflix could become some sort of an entertainment monopoly, or even too dominant, seem a bit overstated given how actively some other very big companies are gearing up to compete in streaming. A Disney boosted by Rupert Murdoch’s entertainment assets will certainly have the resources to compete with Netflix both in terms of scale and international reach. The combined AT&T–Time Warner, if that deal survives the government’s legal challenge, will have more resources to invest in HBO. Apple doesn’t (yet) seem likely to fashion its upcoming TV service as a direct challenge to Netflix, but with an estimated content budget of $1 billion per year, it has already proved it can keep projects away from the company. Amazon has a new executive team respected by Hollywood and as many billions as Jeff Bezos wants to spend, while Facebook and Google (via YouTube) are also stepping up their video efforts. There will almost certainly be fallout as TV continues to evolve from a linear medium to one where content comes on-demand. And Netflix is clearly positioned to be a huge power in this new world order, as evidenced by its ever-rising stock price and subscriber tally. But right now, the average American cable bill is about $100 per month. Even assuming Netflix continues to raise its price (last year, it went up from $9.99 per month to $10.99), consumers will still be able to spend a big chunk of their entertainment budget supporting competing services. “There’s plenty of room for multiple players to be successful in this space,� Sarandos says. “If you love Game of Thrones, you don’t cancel HBO to join Netflix. I don’t know why that dynamic wouldn’t continue to play out with Disney, Apple, Amazon, or anyone else.�

Not that Sarandos thinks competing with Netflix is going to be easy, particularly for old-media companies. “I don’t think it’s in their DNA,� he tells me. “The one thing we’ve been able to do is keep a foot firmly rooted in Silicon Valley and a foot in Hollywood. We don’t jam the tech culture on the entertainment company and vice versa.� By contrast, Sarandos argues, “no studio has been particularly successful with their tech initiatives, and it’s also true that no tech company outside of Netflix has been particularly successful with their entertainment initiatives. That is what’s different about the Netflix story from everyone else. People underestimate the 1,000 engineers in Silicon Valley who make Netflix work every single time you push play.�

In a research letter last year to clients, Sanford Bernstein media analyst Todd Juenger summed up the conflicting viewpoints Wall Street investors — and many in Hollywood — have toward Netflix: “A wise investor once remarked to us, ‘If Jesus were a stock, he’d be Netflix. You either believe or you don’t.’ � Juenger admitted the obvious: Netflix doesn’t actually make any real profits right now because of the billions in debt it’s taking on as it builds up its library. But the analyst, whose bullish predictions regarding Netflix’s stock price have so far been borne out, is confident the company’s investment strategy will pay off as that cycle Sarandos describes unfolds — more subscribers, more content, more people watching more hours of Netflix. Juenger compares Netflix’s trajectory over the next few years to Amazon, which made no money for years and now posts quarterly profits in excess of $1 billion. Wall Street doesn’t always get things right, of course, and as Netflix’s stock price has risen the past year, some skeptics have started speaking up. In March, Citron Research’s Andrew Left, who’s built a career on short-selling stocks, told The Wall Street Journal that “the market has become overenthused by the product and not realistic about the economics.�

Still, Juenger’s theory for why Netflix will succeed seems logical: Subscription video-on-demand services are rapidly replacing linear television, Netflix is the clear leader in the category, and there are still hundreds of millions of potential subscribers to acquire.

But let’s assume that Netflix’s near-term plan plays out much the way things have unfolded since House of Cards launched and its programming portfolio continues to expand exponentially. The company’s strategy of saying “Yes� to talent and creators and expanding into more and more genres probably means its 2023 slate will make its current offerings seem meager in comparison. Disney and other studios have already started the process of removing their content from Netflix’s library, so the company will become even more reliant on originals. And while my visit to Netflix focused on TV production, Sarandos is clearly looking to disrupt the film business, too: With former Universal Pictures executive Scott Stuber leading the features division, movies seemed poised to become an even bigger part of the programming mix.

However, painting Netflix as the villain on the TV landscape might also be giving it too much credit. It’s been clear since NBC put episodes of The Office on iTunes 13 years ago that digital and on-demand were the future of television, and yet media conglomerates such as Comcast and 21st Century Fox opted against sacrificing short-term profits to invest in streaming plays of their own. Save for a handful of exceptions, networks have also been stubbornly slow at changing how they make TV, sticking with outdated development and notes processes that make most networks unattractive to top talent. “Those companies deserve to go out of business,� one longtime agent who’s made millions off those networks tells me.

And some of them will go away, or at least adapt to survive. It’s easy to imagine some very big cable networks shuttering over the next half-decade. Murdoch’s Fox broadcast network is already looking like it will soon be a shadow of its former self, focused on sports and news once it loses sister studio 20th Century Fox TV to either Disney or Comcast. But these changes aren’t only happening because Netflix spends too much money: TV is simply evolving, from a linear, ad-supported business to a subscription model built around making consumers happy — not advertisers. Netflix has had a big head start in shaping this new world, and it figures to be rewarded for being first. This would be nothing new in American media, by the way: NBC and CBS dominated the early era of television mostly because they were first to the market (and already had decades of brand awareness with consumers from their previous iterations as radio networks).

If history is any guide, a dominant Netflix doesn’t automatically equal a TV landscape devoid of choice, for either makers or consumers of television. When NBC was at its Must-See TV peak in the 1980s and ’90s, the network crushed its rivals in the ratings and had its pick of top talent, all of whom wanted to be on the same channel as Cheers, Seinfeld, L.A. Law, and ER. But the other networks still managed to launch hits and make plenty of money even as NBC thrived. While Netflix has one advantage NBC never had — unlimited shelf space, since there are no time slots in streaming — even it can’t afford to hire every good development executive or do a deal with every smart writer with a good idea. Scale makes sense for Netflix, but there’s no reason to think a more boutique approach can’t continue to work for established brands such as FX and HBO — particularly since both of those networks are part of giant conglomerates.

And while it probably won’t happen anytime soon, at some point even Sarandos says Netflix will start pulling back on its content spend, or, at least, stop increasing the budget every year. I ask Sarandos when he’ll know it’s time to hit the brakes. “When we are not growing the subscriber base or hours of engagement per subscriber,� he says. “When you start seeing them plateau, then you say there’s a point of diminishing returns on the continued expansion of the library. But we are not seeing any evidence in either metric yet.� Sarandos also points out that while Netflix users are spending more time watching the service than ever, it’s still “a very small portion� of the overall amount of time we spend on our TVs and cell phones. “There’s tons of growth in user screen time on Netflix,� he says. “If you think about the addressable market for Netflix as being paid TV households, it’s relatively small. Everyone with a phone has a screen and access to the internet. That is our addressable market. The world’s taste, and the world’s time, is what we’re after.�

*This article appears in the June 11, 2018, issue of New York Magazine. Subscribe Now!