Netflix’s ostensible goal is to win the Moment of Truth: when you’re home from work and too tired to do anything but vegetate, are they your first choice? But for investors, this breaks down into two separate, related missions:

Be available to anyone who enjoys passive entertainment and has more than $10/month in disposable income.
Replace continuous and expensive marketing spending with discrete, much-less-expensive content spending.

To be optimistic about Netflix, you have to believe that the greatest movie and TV studio in history was founded by someone who claims his eureka moment was a $40 late fee for renting Apollo 13 on VHS.* Stranger things have happened, but that’s pretty strange.

*The book Netflixed gives a decent summary of how Netflix actual came about. Given the sources the author cultivated, I suspect that the book was originally going to be about the turnaround at Blockbuster, and that the author reworked it and added some more Netflix material once that turnaround failed.

There’s a good reason to think Netflix can, in fact, build the world’s best movie studio. They have a monopoly on the viewing data of 94m people who watch ~1.5 hours of Netflix per day. This is one of the crown jewels of media data. Nielsen ratings cover more viewing time (for now), but broadcasters, studios, and investors all subscribe to them. YouTube covers more online viewing time, but their content leans towards non-exclusive music content at the high view-count end and user-generated content in the long tail. And even if YouTube knows that cute animal content performs well, it’s hard to translate that knowledge into an increase in dog-on-skateboard or cat-in-weird-pose video content. And while granular stats like percent completion and common stopping points aren’t available, raw viewing data is widely distributed.

Since Netflix is using paid content, their viewing data is structured: for a given show or episode, they know who wrote it and who starred in it. It wouldn’t surprise me at all to know that Netflix had scene-level data on actors, and that they tracked which actors predict that the viewer will stop watching, and which predict that they’ll keep bingeing.

The challenge for Netflix, like most other consumer Internet companies, is to grow their information advantage by growing their sample size. They can afford for there be better shows elsewhere — for now. They can lose some of their engineering talent to Facebook and Google — temporarily. They can even tolerate Amazon offering video as a loss-leader — as long as, for everyone in Netflix’s addressable market, there’s at least one Netflix-only show they can’t miss.

Ubiquity

In the last two years, the Netflix story has been a story about broader distribution in two dimensions: more countries to subscribe in, and more platforms to subscribe through. They’ve had massively successful launches in Northern Europe (late ’14) Australia/New Zealand (mid ‘15), Southern Europe and Japan (late ‘15), and everywhere in the world not subject to US sanctions (Jan ‘16).

Their global launch at the beginning of last year didn’t fit the spirit of the previous launches, nor did it fit the letter — literally, they didn’t translate the site into local languages. Netflix Russia was available to anybody in Russia, but if they didn’t speak English they couldn’t read it. So the launch was less transformative than it looked. Netflix wasn’t so much deciding to launch in every market; they were announcing that they’d no longer disclose when they really launched.

You can pick up hints. For example, a few quarters ago they mentioned that they’d localized their Polish and Turkish sites, which had previously been English-only and required an international credit card. More recently, they’ve signed an exclusive deal with one of the most popular actors in Bollywood.

This launch-plus-continuous-soft-launch strategy has worked for them in other markets. In Brazil, for example, Netflix’s success was a combination of:

Investing in Internet infrastructure, both directly by giving ISPs servers, and indirectly by convincing smart TV manufacturers to target the Brazilian market.
Offering better payment options; credit cards aren’t as ubiquitous outside of the developed world.
As a distant third, offering localized content.

This strategy is far from tapped out. Even in the US, Netflix has expanded its availability to set-top boxes (until a few weeks before its launch, Roku was going to be a Netflix-branded product), iOS (late ‘15), Google Play (mid ‘16), and Comcast’s X-1 (late ‘16). Their attitude seems to be: if it has a monitor and audio output, and it’s ever connected to the Internet, you definitely need to be able to watch Netflix on it, and should probably be able to subscribe to Netflix through it.

One of their strategic considerations is clearly that they can go after a slightly bigger market this way. Some people just prefer to watch videos on TV, or on their iPad. But another consideration is that if Netflix is always available on the screen that’s physically closest to you, it’s easier for you to reinforce your Netflix habit. And, as I’ll get to in a later section, your propensity to keep using Netflix once you’ve signed up is the biggest driver of their growth.

Why Originals Work

Netflix would be an easier company to run if the crux of the business was merely to strike biz-dev deals with every single electronics company and digital distribution platform, and then handle the delivery of about a third of all data delivered in North America.

But it can’t be that easy.

Because merely delivering video online is a mature business, at least in the US. Not only that, but Netflix’s biggest competitor is YouTube, which is free. Their biggest competitor for scripted shows is broadcast, which is not only free but doesn’t even require an Internet connnection. Clearly, they’re screwed unless they pair their superior user interface with shows people absolutely have to watch and can’t see anywhere else.

Hence the need for House of Cards, Orange is the New Black, Making a Murderer, The Crown, half a dozen Marvel shows, half a dozen sci-fi originals, Narcos, and what seems like an endless supply of comedy specials.

Netflix as a content delivery business is a mature company. Netflix as a content delivery business plus the movie Girlfriend’s Day is a mature company with a slightly larger addressable market. Their ~50m subscriber base is expanding in small, content-driven increments as they identify new actors and niches and crank out new shows. And that’s something they can do better than literally anyone in the world.

Netflix has the world’s biggest concentration of structured data on exactly what makes a scripted show work. And viewers tend to opt into professional content over user-generated content (it’s not an accident that early YouTube dragged its feet so much on removing South Park episodes and new movies — their gamble was to borrow good content long enough to get momentum with people whose content they could actually buy or host, then pay the bill when it came — this worked beautifully).

So Netflix originals can arbitrage actors’ and producers’ and genres’ perceived popularity compared to their measured popularity, and can also tell who people won’t get bored of. Is the appetite for superhumans in spandex as infinite as Disney thinks? Netflix knows.

It’s worth asking: do people in the movie industry really want to work at a company that can quantify the impact of their casting decisions and tell them the exact net present value of choosing their second-favorite star? Probably: creatives have been arguing with financiers in Hollywood since the beginning, but at least this time the people with money also have data.

What Netflix Tells Investors; What They Could Tell Investors

Netflix’s reported financial statements are a case study in being generous with some kinds of information, and stingy with others. Netflix helpfully tells investors what the contribution profit from each division is, so when you look at their aggregate net income, you can break that into the fairly mature and profitable US business and the growing, money-losing international business. But they won’t tell you exactly where their international signups are — “international” is a term for 95.5% of the world’s population — and, more frustratingly, Netflix won’t reveal churn.

This means that net subscriber adds — the key metric, the only thing that actually moves the stock — is just the tip of the iceberg, and it’s an iceberg in stormy seas. A million net adds might mean two million signups and a million lost users, or three million signups and two million lost users. And that becomes an urgent question when they report numbers like their Q2 results last year, when US subscriber net adds were down 82% from the previous year.

This probably constituted a small percentage drop in gross additions and a small percentage increase in churn, when the two numbers were already fairly close. But management doesn’t break it out, other than at a qualitative level. So, the worse a quarter is, the harder it is to judge how bad things really are. As anyone who followed the Great Financial Crisis knows, when there’s a correlation between bad news and confusing news, prices drop fast. And for a $60bn company, Netflix’s stock is pretty volatile. On average, when NFLX reports earnings, the stock moves about 13%. Priceline, a company of comparable size with comparably tight-lipped investor relations personnel, moves an average of 9%.

What investors really want to know is churn. On average, what are the odds that a paying customer this month will leave next month. This gives us two useful pieces of information:

We’ll know what their actual customer acquisition cost is. (If you look at average marketing spend per net add, it jumps all over the place, especially when net adds are low. But average cost per gross add should be more stable.)
We’ll know the lifetime value of a Netflix customer. Ignoring the time value of money — a popular decision among people who started investing in the last five years — LTV = (monthly subscription)/churn. If people are paying $9.99/month, the most popular price point, and their churn rate is 3%, their LTV is $9.99/.03 = $333. If churn is 5%, LTV is about $200.

You’ll notice that, as churn gets lower, the marginal impact of a small change in churn gets bigger. That’s mathematically obvious, but it’s not instinctively clear to many investors that, in percentage terms, the LTV benefit of going from 2% monthly churn to 1.9% monthly churn is the same as the benefit of going from 5% churn to 4.75%. You’d have to get churn from 5% down to 4.4% to see the same dollar impact on LTV as the 2% to 1.9% swing. (So if you’re wondering how many obscure Korean thrillers or new Marvel shows or quirky comedies Netflix can afford to order, if they’re making loyal users even more loyal the answer is a lot. The mathematics of churn are the best thing for niche artistic interests since the invention of patronage.)

My view is that, over time, originals are increasingly about reducing churn rather than bringing in new viewers. After all, Netflix has much better information about the viewing habits of people who are on the platform. If they’ve signed up 10% of the market in a country, and romantic comedies are huge, does that mean romantic comedies are huge in that country or that of the 10% of people who watch romantic comedies, 100% have signed up for Netflix? But reducing churn now means reducing marketing spend later; in any given quarter, gross adds are probably over 2x net adds (this is the only way to reconcile the jumpy cost per net add number), implying that over half of Netflix’s marketing efforts involve making up for lost customers, not winning new ones.

Content spend has another advantage over marketing: you can amortize it over more people as you grow. As anyone who has discovered an untapped marketing channel knows, it’s very hard to scale up your spending without suddenly paying much, much more for your traffic. But if you think of content costs on a per-user basis, Netflix management has said to expect content costs per user to decline once annual content spend tops $4bn. (It’s at $5bn as of 2016.) So rather than buy users, Netflix can pay less and less money and need to buy fewer users, but show the same total growth rate.

Ultimately, investors play this weird Netflix Kremlinology game, where the really important variables (how much are people watching originals, and what is that doing to churn) don’t get revealed, but if you make assumptions about them then Netflix will give you enough information to get to a price target. This has made Netflix one of the most maddening stocks to trade: while data providers can get you a good read on how things are trending, the fact that the company keeps adding new countries and distribution channels leads to an arms race. And since their reported net adds number is the result of gross adds and churn, and different platforms have different ratios of gross adds to churn, there’s a further layer of guesstimation.

The more time you spend on Netflix, the more apparent it is that the company is a black box. This is for prudent strategic reasons. Local TV companies don’t need to know how many subscribers Netflix has in each country. And Hollywood agents definitely don’t need to know which of the people they represent are starring in huge hit shows. Netflix’s decision to stop disclosing churn is a little more mysterious, since it makes every seasonally slow quarter a white-knuckle experience and doesn’t seem to matter much strategically.

The mission, I understand. The business drivers, I get. The stock, I generally don’t.