In Q2 2023, PARQOR will be focusing on three trends. This essay covers "Media companies have millions of consumer credit cards on file. What are they building for their customers?"
PARQOR identifies a few key trends each fiscal quarter that reveal the most important tensions and seismic shifts in the media marketplace. Must-read stories or market developments are not always obvious from press reports or research analysis, and often require a deeper dive. PARQOR’s analysis questions established ideas and common wisdom, reassesses the moving pieces, and reveals the potential in the media marketplace in 2023.
One of PARQOR’s key trends for Q2 2023 — “Media companies have millions of consumer credit cards on file. What are they building for their customers?” — has popped up in a few conversations in the past week. I received some pushback in conversations on how I have phrased it, as this is not the first time media companies have had millions of credit cards on file. They have had other retail businesses like merchandising or mobile games.
This particular PARQOR trend relies on the premise that with all this consumer data, “they should be able to figure out additional direct to consumer models.” However, the takeaway from recent earnings calls is that these companies have yet to figure out the streaming business model, first.
Disney’s FY Q2 2023 earnings call yesterday reflected this dynamic. The legacy media company is widely celebrated for having figured out streaming reported fewer subscribers and plans to distribute Hulu with Disney+ content in “a one-app experience.” But, Disney’s earnings led commentators like CNBC’s Alex Sherman to argue “the streaming wars are over” and therefore it is “highly unlikely growth will ever return to the levels seen during the pandemic and the early years of mass streaming.”
Something has changed. It is no longer worth asking what additional retail services the media companies may build because they have conceded they are still figuring out their streaming business models. Instead, it is worth asking whether they have misdefined the competitive dynamic as one focused on the math of libraries, when the dynamic is actually focused on the math of content relevance to consumers.
Key Takeaway
What can media companies build to compete with a firehouse of content whose volume is expanding daily? Betting on streaming libraries alone has been a fool’s errand answer. They need to also solve for relevance.
Total words: 2,200
Total time reading: 9 minutes
Defining the problem of volume
The current writers’ strike is protesting the unintended financial, operational, creative and human consequences of the unprecedented scale of “Peak TV” and the streaming era. The implication is that the Hollywood system is no longer working for creatives, and therefore media companies need to refine, rethink and readjust their approach to streaming.
It is ultimately questioning the “Peak TV” demand that streaming created — 599 English-language shows in 2022, up 42% from 420 in 2015 when FX Chairman John Landgraf coined the term and up 2.85x since 2009 when FX first started tracking the total. That has resulted in writing rooms with fewer writers, shorter seasons, and shrinking budgets. The streaming business model seems unsustainable for everyone in the ecosystem, including Netflix, but especially for the writers.
Volume, in this sense, has always been framed in terms of library titles. Every streaming service has invested billions in original content —a projected $26.5 billion in 2023, up 14% from $23.2 billion in 2022, according to recent research from Ampere Analysis — because larger libraries have been perceived as a competitive advantage. Paramount Global markets Paramount+ as “a mountain of entertainment” with 500+ movies and 400+ shows.
But the problem of volume as framed by writers’ strike is more specifically a problem related to Netflix’s model. As Jaclyn Moore, an executive producer and writer for “Queer as Folk” on Peacock and “Dear White People” on Netflix, told The Los Angeles Times: “Netflix in a lot of ways has upended the business model, and broken it in fundamental ways.”
Netflix, volume & relevance
When Netflix started producing original content in 2013 — launching its ambitions with “House of Cards” — it had less than 20% of the subscriber base (41,3 million) than it has now (232.5 million). 10 years later, Kasey Moore of What’s On Netflix estimated that over 50% of its U.S. library in August 2022 — 3,104 titles of 6,206 individual titles on the service — were Netflix Originals.
In other words, Netflix’s total library of originals is over 5x the total number of new shows produced in 2022. So, scale is one advantage for Netflix.
Another advantage is Netflix’s focus on “content people love”, as it tells investors on its investor relations site: “People's tastes are very broad, even in a single market. The internet allows us to offer a wide variety, and to have our user interface quickly learn and make recommendations based upon individual users' tastes.”
Netflix’s interface is the key factor here: “By personalizing promotion of the right content to the right member, we have a large opportunity to promote our original content, one that's effectively unlimited in duration.” The point is not that they have an enormous library of content as a competitive advantage, but rather because their user interface and the technology behind it reduce the scale of their originals into personalized offerings. They are constantly testing content recommendations, photos and videos within individual tiles for the series, and text descriptions.
That means that Netflix’s software is not distributing 3,104 individual titles. Rather, it is figuring out how to take elements of each of those titles to make them relevant to its subscribers.
Take thumbnail images: each show has as many as 30 different thumbnail images. The images that get clicked on the most surface to the top and are then used more frequently. The images are also customized for viewers. For example, after the success of “Queen’s Gambit” on Netflix, viewers who had watched the show would be served a thumbnail for the original series “Peaky Blinders” with “Queen’s Gambit” actress Anya Taylor-Joy as the image. However, that series was built around Cillian Murphy, star of the upcoming Christopher Nolan movie “Oppenheimer”. Taylor-Joy did not appear until four seasons and 24 episodes into the series, but the success of “Queen’s Gambit” helped Netflix to make an unrelated show ("Peaky Blinders") on an unrelated topic (Northern Ireland) more relevant to subscribers who watched a show about an American chess prodigy.
Thirty different thumbnail images for over 3,000 titles offer nearly 100,000 thumbnails to test with consumers. The permutations and combinations of those tests make the math of relevance exponentially large, especially when you factor in tile location and titles suggested in the same row. This all enables Netflix more “at bats” with the consumer to get them to click on content to watch. Throw in video and text descriptions and the permutations of the offerings seem endless.
So Netflix can have both a large library and an approach to relevance that makes its library “larger” by algorithmically finding ways to make titles more relevant to consumers. Mathematically speaking, 3,104 original TV series are a fraction of the versions of these shows that users see of these series in Netflix’s interface. Different thumbnails for the same show are effectively selling different versions of the same show.
The medium is a firehose
According to Nielsen, Netflix has 7% less share of consumption (7.3%) than YouTube (7.8%) on Connected TVs, the most valuable real estate for an ad-supported business (YouTube) and an aspiring ad-supported business (Netflix).
As I wrote last month, “today, more than 500 hours of content are uploaded to YouTube every minute. Over 50% of content consumed on YouTube is creator content, and there are over 2 million creators in its Partner Program which launched in 2007.” CEO Susan Wojcicki has said more than 50% of content consumed on YouTube is creator content.
Also, The Information’s Sahil Patel reported last week that YouTube’s “internal data indicate that close to 45% of overall YouTube viewing in the U.S. today is happening on TV screens, according to people familiar with the matter, compared with well below 30% in 2020.” Nielsen has reported YouTube reaches 135 million connected TV users across the U.S., and Google has shared that YouTube Shorts get more than 50 billion views per day.
On top of that, its creator model seems to have infinite supply: an observation I have heard casually repeated about creator burnout on the Colin and Samir podcast estimates that if 1,000 YouTube creators burn out, another 1,000 YouTube creators are in place and ready to take their places.
Like Netflix, YouTube’s interface serves personalized recommendations. As I wrote last October, “users of the YouTube CTV app are recommended content based on a recommendation algorithm, they are served ads by an algorithm, and their comments are moderated by an algorithm.” The only difference between YouTube and Netflix in terms of relevance is that the uploaders have control over the thumbnails, video content and text. YouTube’s algorithms may play a role in recommending thumbnails, but the back end is not designed to A/B test them.
So, YouTube’s model is effectively a firehose of content regulated by relevance algorithms.
AI enters the firehose
Content created by artificial intelligence (AI) will only add to that firehose and increase the volume of content produced and published to platforms like YouTube. I argued last October that “the future business models for videos produced by AI engines like stable diffusion and DALL-E are going to look a lot like the creator economy.” That dynamic has been playing out at Spotify lately: as of last October, 100,000 tracks are being “added to music platforms every day” according to Universal Music Group CEO and Chairman, Sir Lucian Grainge. He also argued “this vast volume of music, plus additional “associated content” on social platforms, is making it increasingly difficult for artists to break through to a substantial audience online.”
On Tuesday, Spotify announced that it had taken action against AI-generated songs on its platforms, and removed thousands of songs from artificial intelligence music start-up Boomy. According to The Financial Times’ Anna Nicolau, Boomy launched two years ago. It “allows users to choose various styles or descriptors, such as “rap beats” or “rainy nights”, to create a machine-generated track. Users can then release the music to streaming services, where they will generate royalty payments. California-based Boomy says its users have created more than 14mn songs.”
Even though Boomy is not a video service producing original TV shows, AI is very much a worry for TV writers because it generates exponentially more content in a short amount of time. They may have an AI-generated version of the TV show “Seinfeld” that took over Twitch in February on their minds as a worrying example. The channel, WatchMeForever, uses generative artificial intelligence to create both the audio and visuals of an infinite stream of Seinfeld-esque content in low-fi graphics online.
With over 10 million active channels on Twitch, and 100,000 pieces of AI-generated content being uploaded to platforms, the math gets scary for writers, and fast. So, writers are asking the studios “for guardrails against being replaced by A.I., having their work used to train A.I. or being hired to punch up A.I.-generated scripts at a fraction of their former pay rates.” In other words, AI offers competition that is faster and cheaper and perhaps sometimes better than what Hollywood creatives have been compensated to deliver. Creatives need guarantees of protection from media companies who inevitably will need to find cheaper content after having misallocated billions of dollars in capital to building content libraries.
What to build?
The problem of volume raises a simple but difficult question for media companies: What can they build to compete with a firehouse of content whose volume is expanding daily?
They have taken on billions in debt and lost billions of dollars annually in operating income to drive a streaming business that simply cannot compete on volume because the competition is exponentially large due to YouTube’s and Netflix’s models. AI presents the threat of an exponentially larger set of competitors. That means, betting on streaming libraries alone is a fool’s errand. And, along the lines of Alex Sherman’s argument above, I think that is what we are seeing being proven out.
But Netflix’s focus on relevance may be the most important detail here. It answers the question I asked in March of “Why Don't We See More Crunchyrolls?” To remind you:
Crunchyroll connects anime and manga fans across 200+ countries and territories with the content and experiences they love. In addition to free ad-supported and subscription premium content, Crunchyroll serves the anime community across events, theatrical, games, consumer products, collectibles, and manga publishing.
Crunchyroll hyper-serves 10 million subscribers who are fans of a genre with a business model built upon the assumption that those subscribers will always want anime and manga content, but they will not always want to stream it. In other words, Crunchyroll competes in streaming by not always competing in streaming but always focusing on relevance. It does so by delivering and monetizing content in the channels that match the affinities of the most passionate and engaged consumers, and not by aiming broader by investing in “general entertainment”.
It is a much smaller business than Netflix or Disney+. But, as a portfolio of services and offerings for passionate fans of a genre, it has a foundation for the future that few seem to be questioning (in large part because Sony has not bet the future of its distribution on Crunchyroll). In that light, the question of what to build is less about technology and more about a portfolio of services that monetize one user in ways that are relevant to that user.
This all would suggest PARQOR’s key trend should be rephrased as “Media companies have millions of consumer credit cards on file. How are they solving for delivering more relevant content and services to their customers?”