Reading "The Streaming Book" by @ballmatthew

What We’ve Learned About Content (Distribution)

Just as the access wave clarifies access-related business practices, the content wave reveals how best to produce and market content—as well as what matters when doing so.

The most obvious change is when hit content premieres and airs. For decades, almost every series launched in September and ended in May. In the interim, holidays would be taken off (Thanksgiving, most of December, mid-March), as would any release that would overlap with major sporting events (March Madness) or national ones (Election Day). Networks would also release (and market) their “best” episodes during a roughly 3-week period in November, February, and May known as “sweeps.”

This model was shaped by limitations. For example, TV was “live only,” and so airing shows at a time in which a greater-than-normal share of the audience was likely to be unavailable was wasteful. And given that the average show had 20–24 episodes per calendar year, networks could skip more than half all weeks in a year.

Another driver of seasonal scheduling came from TV’s ad-based business model. Every 30 minute scheduling block afforded roughly eight minutes of ads, or 16 ad slots – and because of linear distribution, a given slot could air only a single ad (i.e. there was no targeting, no different ads for a 16 year-old man and 75 year-old female retiree). This made it essential that a network negotiated the highest possible price for this slot, which in turn necessitated auction-style pressure. In order to maximize their revenues and make the largely manual slot-by-slot ad sales process more manageable, networks began to concentrate their ad sales efforts into batches.

The major sales effort happened every May (the “Upfronts”), when networks would invite top advertisers to in person events and advance screenings of the fall’s premieres. With any luck, networks would book as much as 75% of their revenue during the upfronts, with blue-chip advertisers frequently committing 50–90% of their 12-month TV ad spend. Notably, the Upfronts were timed to coincide with most show’s season finales, during which press attention (and ratings) typically peaked.

Much of a network’s remaining inventory would then be sold in batches around Nielsen’s “sweeps” period, which was effectively a popularity check for the networks’ shows. To boost sweeps viewership, networks would typically air attention-getting gambits like cross-over episodes (the Cheers crew shows up on Frasier!) or recruit some big-deal guest stars (Brad Pitt on Friends!) or resort to other gimmicks (the disgusting phenomenon of the “lesbian kiss episode” was tailor-made to serve this function.

The streaming era has reset historical perspectives on TV release models. To begin with, many streaming series are binge-released, thereby making the idea of “weeks off” irrelevant. But even series that adhere to weekly releases will rarely, if ever, pause. Sure, an episode might drop on a holiday weekend, or election night, but streamers know that interested viewers will catch up a day or two later. Sometimes a network might even pull up a single episode’s release by a day or two – as HBO did with The Last of Us’ fifth episode, which would have aired during the Super Bowl LVII – an unthinkable move in the linear era. After all, most viewers don’t watch a show the moment it comes out. Ads, meanwhile, are bought and displayed programmatically—even if advertisers make large dollar commitments at the upfronts—meaning that the “best” episodes air, well, whenever. More importantly, streaming services need fresh hit content as much in one month as another because each month is a new billing period for month-to-month-subscribers.

And so over time, we have seen a progressive smoothening of the release models. There are still spikes (partly because a third of TV series still premiere on linear), but every year, demand of TV content evens out because series releases occur year-round. In fact, many franchises (e.g., Marvel) are themselves year-round franchises that launch new entries year-round. In 2021, Disney+ launched MCU series in January, March, June, August, and November, with a Star Wars TV series dropping on December 29. Both HBO and Amazon chose the period right before Labor Day to air the debut episodes of their most hotly anticipated shows of the year, referring to Game of Thrones: House of the Dragon and The Lord of the Rings: Rings of Power, respectively. A decade ago, such a move would have been unimaginable even for a throwaway series.

While there are no longer technological reasons to release content on a weekly basis, such a model is probably preferred by audiences, as well as better for the show itself and the service which airs it. At the most extreme, binge releases mean a user might sign up for a single month in order to watch one entire series in a day, weekend, week, or even month or two, then cancel. But even when it comes to long-running subscribers, binge releases lead many to chew through a service’s best content in a short period. Whether subscribers do this to save money or not, the end result is a shortage of top content to watch a month later. And this matters when the price for these months is the same, and when the impact is applied to hundreds of millions.

None of the above means consumers wouldn’t, in the moment, prefer to binge a show. And it’s often smart to give customers what they want—especially if it leads them to use your service more in the moment, rather than hope they return in a week to do so. Yet the Marshmallow Test is a thing, as are diminishing marginal returns. Consider your favorite chocolate bar: you’ll enjoy eating one each week more than eating two back-to-back. Same for consuming a sizzling steak, attending a football game, or going to the beach.

Essential to understanding the role of weekly releases is the importance of network effects. We often talk about Metcalfe’s law in the economics of video games because that media category is so inherently interactive, social, and multi-user. However, the phenomenon applies almost as well to books, music, TV, film, content categories that are non-interactive and are often consumed solo. The Bachelor is better as the number of your friends watching it increases, better still if they’ve watching live with you, even more so when you’re all on the same couch, and best of all in a bar that is hosting an event for it. The press furthers these effects by both promoting a series for longer, while also joining in the experience of watching it, and thereby enhancing it, too, per Metcalfe’s Law.

For these reasons, we can observe an ever-growing focus on weekly releases. Today, Netflix is the only real holdout on the model, but it continues to split seasons of flagship series into two batches that are released a few months (Ozark, Grace and Frankie, BoJack Horseman) or barely a single month (Stranger Things, You) or, occasionally, as little as a week (Arcane) or two (The Sandman) apart. Netflix also pursues this strategy more frequently than ever. Prime Video spent years releasing all of its series via binge drops, but now staggers all of its best series out on a weekly basis. Apple TV+ has always taken this approach.

As a corollary, we should note that a streaming service’s subscriber base also enables its content to be “better.” Tiger King is unlikely to have popped on any U.S. streaming service in 2020 other than Netflix, which boasted 4–5 times the users and a multiple of the usage per user, too. A water cooler needs to be pretty packed to convince you to watch Joe Exotic – especially if you’re not yet a subscriber to the service that has his series. This theory is constantly reiterated. How many people know that in 2019, Showtime had a drama series about Roger Ailes at Fox starring Russell Crowe, Seth MacFarlane, Sienna Miller and Naomi Watts, or, a year later, a Civil War epic starring Ethan Hawke, plus a James Comey drama series starring Jeff Daniels, Holly Hunter, and Jonathan Banks, or that its Kevin Bacon-starring cop drama produced by Ben Affleck recently wrapped up its third season? The Queen’s Gambit was superb, but its commercial success and critical acclaim (11 Emmys, including the first “Outstanding Limited Series” win for a streamer) would have been capped much lower on another streamer. Put another way, not every streamer can achieve the same high from the same content, and certain major players are able to get more out of a lesser show, too. This is a virtuous cycle for the Goliaths and a vicious one for those would-be Davids. If a service needs outstanding content to grow, but the service’s modest scale impedes the success of that content, while the scale of larger services means even their middling series can thrive… that’s a hard cycle to escape. Which is why we continue to see “SKU collapse in SVOD,” with Paramount finally merging Showtime into Paramount+.

The impact of reach on the popularity of (and thus return on) a title is best shown through Nielsen's streaming data (this dataset is unfortunately limited to a few years). Between 2019-2022, Netflix produced a shocking 45% of all original series released by streaming services. It also averaged about 30% of the total supply of licensed/second-run/library titles available on streaming services (data here is choppy). And last chapter, were viewed that Netflix holds a roughly 40% share of total demand for a streaming service.

In contrast to the figures above, Netflix maintains an average 75% share of minutes watched in Nielsen’s Top 10 Streaming Originals charts each week), 65% of the most watched films, and 60% of licensed content viewing. This sum is down since early 2020, but this was inevitable given Disney+ and Apple TV+ launched in Q4 2019 with only a few originals each (Apple had no library at all) and have scaled up considerably since then, while Peacock and HBO Max had yet to debut, with Paramount+ was still a year from its rebrand and Roku a year from its originals programming.

Netflix’s outperformance is a powerful moat. As the access era ended, competition has grown, Netflix’s library has shrunk, and while its originals have grown in number, their share of market has plummeted from the early and mid-2010s – and Netflix’s price has more than doubled! Still, the service remains nearly as popular as ever and the share Netflix has lost is heavily fragmented. Most people will subscribe to and keep the service with the most popular content – especially if the alternatives are manifold. And in turn, this choice helps Netflix invest in more content for that subscriber, and then generate returns on that content that another service would struggle to achieve.

The role of network effects also explains why the arc of scripted and unscripted content in the “over the top era” differs. In the linear days, the average “fan” saw only one of every three episodes when they premiered because it wasn’t always possible to at, say, 8 p.m. on Thursday every single week. And for a hit show, the vast majority of lifetime consumption would happen via syndication—reruns—with individual episodes viewed in what was effectively random order. As a result, a scripted series could not be very complex, with most shows limiting the consequences of anything that happened in a given episode, to that given episode, lest those who missed it struggle to follow along in the future. But now, anyone who wants to watch every episode of a show (and do so several times over) can do so, no matter their schedule. This has allowed scripted television to become far more sophisticated without shrinking the total addressable market.

However, streaming services haven’t figured out how to make unscripted or reality series better when delivered over the top. In fact, they’re generally worse, as streamers have smaller reach than broadcast networks as well as less concentrated consumption. Interactivity is obviously the answer, but relatively few tests have occurred, and so the genre remains mostly unchanged. To this end, it’s worth stressing that the biggest innovation in the category’s history is more than twenty years old, Pop/American Idol, and it’s proof of how potent interactivity can be when it’s done right. After all, the show barely offered much of it (a highly diffuse vote on a single poll each week) in a highly asynchronous form (the input vote and output result were separated by more than 20 hours). For six years, the series was the biggest in the United States, often taking the #1 and #2 slots for its Tuesday (vote) and Wednesday (results) episodes.

Streaming may not have fundamentally changed scripted or unscripted content, but it has led to one revolution: foreign-language content. In the pre-streaming era, consumption of non-English series was too low to sustain a major cable channel. The challenges here were many and intertwined. For example, the network could air only “one” copy of a foreign-language title at once – the unmodified foreign-language original, a subtitled copy in the original language, or a dubbed version. And in both the subtitled and subbed version, only one language could be supported – it was English, or French, or Germany, or Telugu, etc. This made it difficult to air any content other than English, spoken by 80% of Americans, or Spanish, 14% (the third most popular is Chinese, which is spoken by 1.1% Americans when combining Mandarin, Cantonese, Hokkien and all other variety). And if linear distribution made it difficult to accumulate audiences years into a series’ run – well, just imagine how hard it was to attract American audiences to Season N of a Korean title! Linear-only delivery also made it hard to monetize what few English viewers did accumulate, too. Advertisers would buy an ad slot on Telemundo to reach Hispanic/Latinx/Mexican viewers, rather than Caucasian Midwesterners, and would rarely if ever subtitle these ads, too. As such, these viewers drove little-to-no value. A final pair of challenges stemmed from the limited foreign operations and insights of the “major” cable networks. Whether the network was ABC or FX, neither was developing content abroad, nor did they have the expertise to pick which foreign titles to license, or how to market them. And for the most part, the production quality (i.e. budget) expected by English-language American audiences was not available abroad, meaning an import would feel even more out of place than usual.

The advent of streaming quickly solved the bottlenecks above. A streamed title could be distributed at all times, in all supported languages and versions, thereby ensuring every nearly subscriber could join a series at all times. Subscription economics and dynamic ad targeting also meant that all foreign-language viewing was monetizable. And most of the major streamers, from Netflix to Amazon, Disney+ and Paramount+, quickly stood up foreign content hubs which both licensed and produced non-English content, and often at American-style production budgets. Indeed, success abroad became one of their top priorities - and eventually, comprised the majority of spending growth. The result has been extraordinary.

Not only has aggregate demand for streaming content grown over the last decade, but non-English language titles have grown their share of this demand by 350%. In 2016, the category comprised 2% of total demand; in Q1 2023, it was 95%. This demand has also diversified, too. In 2015, half of non-English language demand was for Japanese content (mostly anime), with Japanese and Korean making up a combined 75%. Today, Japanese-language content is less than25% of total demand, with the two. Hindi and French have grown to 15% combined, while “All Other” has grown from 15% to 21%.

Although we're watching more foreign television than ever, you might be surprised to find out how much we continue to use the television itself. Yes, we do watch more premium video on small handheld devices than ever before. However, 2018 data released by Netflix (at which point the service had over 125MM subscriptions and least 450MM users) nevertheless demonstrates the enduring supremacy of and preference for the TV set.

Out of the average 100 sign-ups to Netflix globally, the largest share come from laptop/desktop/Mac devices, which are collectively responsible for 40 sign-ups. The next largest segment is mobile (smartphone and tablet), with 35. The smallest share, 25, come from TV. This makes sense as TV devices are typically the slowest/hardest to sign-up on and least used overall by share of time per day.

Actual engagement looks different. In the first month that these subscriptions are used, TV’s share surges from 25% to nearly 50%,while mobile shrinks from 35% to 25% and PC+Mac from 40% to 28%.More fascinating is what happens as subscribers continue using Netflix. By month six, TV is now 70% of time (versus 25% sign-ups and 48% in month 1) and mobile shrinks to 15% (versus 35% and 25%). While the specific figures differ due to varying broadband and TV penetration rates, this general trend holds globally.

This evolution has two drivers. First, as Netflix subscribers grow their overall viewing hours, most of this growth happens on TV. As a result, the TV numerator and viewing denominator both grow, but the numerator grows more in percentage terms. Concurrently, the TV screen also steals net hours from the smartphone. This reflect a simple truth: the living-room and its TV is the best place to watch most content. We will use smartphones when we must –on the subway, in our bedroom’s when we’re tired or the family TV is in use –but we prefer to watch this content on the big screen. And so, as subscribers get more used to a service, or fall more in love with a given title, they shift consumption. Mobile is a key acquisition channel and engagement booster, but not where the services are primarily rendered.

Note, too, the implications for “mobile-only SKUs”. When Netflix cut its pricing in India for a mobile-only plan, many argued this as a show of weakness or that the company had originally overpriced its service. However, the data listed above tells a different story. If the majority of consumption happens on the living room screen, and the most valued consumption does too, then the standard SKU is overcharging those who can’t often (due to broadband) – or ever – use the service on the big screen. Put another way, the mobile-only SKU is right-sizing pricing and value.

It was never debated whether sports would be live when delivered via OTT services (again, this was mostly because live sports have great network effects; few sports fans watch timeshifted games they are not able to see live, even when they don’t know the outcome). But the big questions were when streaming services would have the economics and technology to support live sports, and when households would have the broadband and inclination to watch them in that form. That time has arrived. Today, there are four categories of sports rights by delivery. First, signature sporting events that used to be linear-only but are not only available through streaming services (e.g., Amazon’s Thursday Night Football or YouTube’s NFL Sunday Night Ticket). Second, those that are also available via OTT, such as those of the PGA Tour, which Paramount’s linear networks and streaming service both carry live. Third, those which are only available OT, but did not traditionally see linear distribution, such as the Premiership Football on Peacock and the Champions League on Paramount+. The fourth category is games that remain pay TV–only, such as March Madness, but this collective is shrinking and can, without exception, be watched through virtual MVPDs such as Hulu with Live TV or YouTube TV. These services aren’t as reliable as the “old thing.” but it’s no longer an impediment.

Although the digital future of sports rights was never in question, many wondered whether the price of these rights would continue to explode. In particular, analysts wondered how traditional (i.e., Hollywood) buyers would be able to afford to pay more for sports rights as the pay-TV bundle declines, and whether the vertically integrated tech giants would be willing to spend their billions on said rights. Time has shown that prices will continue to go up.

Core to this is recognizing that the major sports packages have three attributes that are individually rare and collectively unique. First, they offer large volumes of content (in many cases, hundreds of games). Second, the ratings for this content are highly predictable (somewhere around +/-3% year over year). Third, audience appetite for sports content is extraordinary (so deep it often lasts a fan’s entire life). Compare a top sports package to something like the Marvel Cinematic Universe, one of the most valuable media properties in history. The MCU is a multi-generational franchise (#3), both in terms of reaching multiple generations and being loved by the same fans as they progress through multiple life stages. But while the MCU also produces a high volume of programming each year—nearly 20 hours in 2022 across three feature films, three series, and two specials—it amounts to less than the NFL on CBS in any given week. And although the MCU has a high floor when it comes to its theatrical performance, the specific returns are still highly variable. Some films top out at $650MM, while many cross $1B or even $1.5B. Due to these many attributes, sports rights have historically been used to launch new entertainment platforms, from Fox to DirecTV and now Amazon (NFL), Apple (MLB), and YouTube (NFL). As long as the streaming wars remain unsettled, and there’s always a “new thing” (Wave #3), the top sports package will continue to appreciate.

As many had predicted, theatrical windows eventually collapsed in the streaming era. But as with binge release strategies, which dominated streaming in the 2010s, this trend eventually reversed course. During the pandemic, many titles shifted to day-and-date SVOD and theatrical, while Universal agreed to a 17-day window. A year later, most studios pushed to 30, and now Paramount and Disney lead the charge at a 45 day hold. Much has been said about Tom Cruise’s push for the window, but in May 2021, Marvel Studios came out in full force with a three-minute trailer titled “Marvel Celebrates the Movies,” complete with maudlin voiceover read by Stan Lee: “The world may change and evolve, but the one thing that will never change: we’re all part of one big family. That man next to you, he’s your brother. That woman over there, she’s your sister.” video showed off Marvel’s next three years of films, finishing with the statement “See you at the movies.”

Theatrical windows are important even when they’re short. They help to drive monoculture and signal value, as well as the possibility of a shared experience that, again, we all know makes the content “better.” We laugh at a comedy more in the theater than we do at home, just as we do when watching with someone at home than when we’re alone. At the same time, theatrical releases are costly and risky. Marketing a film typically adds 50% more to its total budget but sometimes fails to drive audiences to the opening weekend. If consumers don’t like the film in the opening weekend, sales can quickly dry up. As such, the window typically makes sense only for films that (1) have wide potential appeal; (2) have a high audience floor; (3) are likely to be loved by audiences; and (4) audiences feel they need to see right away in order to participate in pop or local culture. This doesn’t fit most titles, which is why blockbuster IP now dominates the channel. And in success, these titles generate hundreds of millions in profits.

It was argued that these blockbusters should go direct-to-streaming specifically because they’re so beloved. The logic here was always flimsy. Yes, putting signature hits exclusively on SVOD does make the service more attractive. But it stands to reason that those who aren’t excited to see something in theater and don’t already have a given SVOD service (which is doubtless full of other entries in the same blockbuster franchise) aren’t going to sign up en masse for the service simply because it has that title a little bit earlier. And if they do sign up, it’s likely they won’t stay for long. While the move should help increase the price that current subscribers are willing to pay, they already make these payments at the theater, while also benefiting from the community experience the theater provides and driving additional (and costless) press for the title. Even worse, the service would effectively be raising the price for those who do not value day-and-date theatrical releases. In general, media businesses benefit from customer discrimination because love for content varies enormously, as does the ability for individual users to spend; flattening this out makes little sense.

At the same time, all strategies require context. Each SVOD operator and Hollywood studio sits in a different competition position, with different IP, and different needs. As such, we should not expect every player to have the identical strategy for their entire slates and at all times, let alone believe that such a thing would be optimal. Disney, for example, has the second largest SVOD service globally, and the most profitable and reliable film studio, too. In 2019, Disney was responsible for 8 of the 10 most profitable films of the year – including two of the five most profitable films ever – despite the fact it only released 10 films. Only one of its films lost money. Accordingly, prioritizing theatrical channel makes far more sense to the company than it does for Paramount, whose Paramount+ is a laggard in streaming and Paramount Pictures spent most of the last decade losing money.

Streaming was supposed to change more than just how we released content - it was also supposed to make us smarter about how we did so, and what we made in the first place. So that's up next.