Twitter and Subscriptions
Creator subscriptions, not platform subscriptions, are the future
Twitter put a job posting to work on a “subscription platform” and their stock jumped around 10%. I am contractually obligated to write a newsletter about any major content subscription play so here we go.
Twitter pursuing creator subscriptions could be an effective way to monetize their distribution network and could kneecap the growth of Substack. If Twitter pursues a platform subscription like the Wall Street types think, I will do “something stupid,” to be determined by Twitter poll. Instead, I outline what a creator subscription model would look like and why I think that’s what they are trying to build.
This is a relatively long post, if you’re familiar with platform and creator subscriptions, feel free to skip to the last section, with predictions about Twitter.
The rumors
The hullabaloo is based on this career listing, for a senior full-stack software engineer. Somebody panic deleted it before putting it back up, which means that it is definitely real.
Here is a previously unreported but up since May 2020 job description for a frontend software engineer on the Gryphon team. That listing describes the subscription team as building towards a consumption product. That term could mean any one of a variety of things, but my gut is that it means a product which consumers consume content through.
Exclusive, exclusive, you’re welcome for the research, Wired.
Another free clue: the first job posting specifically mentions working with the Twitter.com team -- if Twitter offered subscriptions in their app, they would have to pay the App Store tax. That tax is palatable for a platform subscription (it’s just the platform’s money), but it would be unworkable for content subscriptions, where the payment is passed onto a creator. You can see the impact in action: Substack doesn’t have an app, and Patreon’s app kicks you to web if you try to subscribe, according to App Store reviews (that pattern is almost certainly a violation of Apple rules too).
Then again, the only industry lazier than tech reporting is tech banking research. Via Bloomberg:
Mark Zgutowicz, an analyst at Rosenblatt Securities, said Twitter is “highly unlikely” to consider paid subscription tiers for its service. Instead, there could be a subscription tier “for data and analytics that its power users may consider.” The initial total market may be fewer than 10 million users, he said.
Sometimes I think research analysts only charge for their research out of embarrassment, so that fewer people can see them be wrong. For one, Twitter already has a data and analytics product. For another, the “TweetDeck on steroids” (aka platform subscription) model hasn’t been discussed since 2017, and for good reason. Platform subscriptions are a subscription to the platform (Twitter) that generally offer some combination of ad-free browsing, exclusive content (often funded by the platform), or other platform-specific benefits.
Let me outline why platform subscriptions don’t make sense for Twitter, the platform, or for actual content creators on Twitter.
Platform subscriptions can’t coexist with ads
Selling advertising is a hell of a drug. It’s hard to wean yourself off, and it’s honestly pretty effective at making money. Platform subscriptions have the core problem of adverse selection: the people who can afford to pay your monthly subscription rate are also probably the ones who are most valuable to advertisers.
Also, internally, you’ll have two teams with competing incentives, which ends in one team’s PM quitting in disgust and the engineers getting Below Expectations ratings. The better the ads team does at monetizing customers, the higher the breakeven price point for a subscription and fewer conversions; the more users who upgrade to premium, the fewer impressions the ads team has to train models off of, and so on.
There are some examples of ad-supported platforms which do offer premium ads-free versions:
Hulu is a great example of how hard it is to wean yourself off of ads - their ad-supported plans generate around $10 per month per customer, making it more profitable for them to have customers on the $6 ad-supported plan than the $12 per month premium plan. It’s a hard business but one where you get better at scale - better algorithms, more ad inventory to choose from, and higher prices.
Youtube Premium is an ad-free version of YouTube, but it’s a shadow of its former self, as the “premium subscriber-only content” is available with ads now.
Reddit’s ad-free Premium offering is a reasonably close analogy to a Twitter Premium, but Reddit’s ads are really bad, so it’s a lot easier for them to give up the revenue stream. Reddit also doesn’t seem to mind a chaotic development process; for them, “competing incentives” is just another Tuesday.
It’s very hard to imagine a world in which Twitter throws away the last three to four-ish years of product development, modernizing their ads platform, in a pivot to subscriptions, right when that advertising focus is starting to work. Twitter now considers that “more than 90% of the projects planned in ads serving can be executed” and “online ranking experiments in the ads serving space are now being run 50% quicker.”
Twitter doesn’t break out their CPM and CTR rates, but note in their most recent 10-Q that CTR’s improved year-over-year across most ad types. Obviously the coronavirus pandemic negatively impacted overall ad spend and as a result CPM, but improved CTR across ad types is what you expect to see as a result of improved ad serving infrastructure and, in particular, improved experimentation and iteration speed.
Platform subscriptions don’t help content producers
My central thesis behind the rise of content subscriptions is that people are willing to pay for high quality, differentiated content to support the creators. This is not a unique take, but I think it’s important to stress the connection between reader and creator -- when I enjoy my Twitter feed (which I usually don’t), it’s because of individual posters, not the “website.” A lot of the analysis around subscriptions is around whether or not they offer utility to the subscriber, and “supporting a $25 billion market cap corporation” is a lot less appealing than supporting an independent content creator.
Medium is a reasonable example for what a platform subscription on Twitter would look like. It doesn’t work.
Paying $5 per month and spreading that out over a number of creators doesn’t result in much money going towards creators.
Creators get paid in proportion to views on their content, so now they’ve recreated the same clickbait incentives as the rest of the internet.
No matter how much you love a writer, you can’t send them more than $5 per month, and that’s before the platform’s cut.
Remember claps? It’s all about reading time now, like every other attention-based platform.
I’m also convinced that Medium has been trying to sell themselves to Twitter for years and I’m pretty impressed that Twitter has consistently said no. It’s telling that Medium is building a new newsletter product; it’s an effective way to stay engaged with an audience.
Substack and Patreon have filled this niche well for many creators: build an audience and distribute your work on Twitter, take payments elsewhere. Why should Twitter let anyone else in the space? Any middleman is inefficient; offer content monetization along with the distribution.
Why are content subscriptions hard?
I suspect that Twitter has been slow to roll out content subscriptions because they need to reckon with the tradeoff between creators and the platform, with respect to who owns the creator-reader relationship. There are different norms for what users see in feed, and these create different expectations around that relationship.
(this took like 20 minutes to make because Facebook’s brand files wouldn’t import into Photoshop)
Simplistically, we might consider the different platforms:
Facebook core app: weak creator-reader relationship. Organic engagement has significantly decreased over the last few years; personally, most of the posts I see come from meme pages. Even professional outlets which invested heavily on Facebook were burned and it seems very hard for them to sustain relationships.
Twitter: medium creator-reader relationship. There is an algorithmic feed (which has improved dramatically over the last few years), but it feels like creators can still easily reach their readers. I am certainly in a filter bubble but it feels like Twitter is the best available resource to build a network with today (c.f. Nathan Tankus).
Substack: strong creator-reader relationship. There is no algorithm or feed, and thus virtually no distribution - you’re on your own with finding readers. At the same time, there is no competition for readers, except in the inbox.
I don’t think any of these models are “Pareto optimal,” but they are all explicit design choices. A key reason why Substack has been successful is because they are very explicit that writers own their readers and the relationship, and they should never fear that the platform will promote other writers to the same set of readers. This is a great response to a platform like Facebook, with their history of burning content creators, but Twitter could be the platform which marries distribution with monetization effectively enough to win out.
Why Twitter?
The dirty secret of most creator platforms is that they don’t monetize effectively well for most creators; the top writers on Substack, Patreon, Medium, etc make pretty decent money, but the revenue drops off dramatically beyond the top few creators and most barely make anything. Substack sends writers a recurring series on how different authors built their audiences, and Twitter features prominently in nearly all of the stories. So if Twitter is a platform to build a brand, why not also be a platform to monetize that brand?
Established content creators are already well served by Substack/Patreon, as they already have distribution networks and the lower fees would add up, especially at scale. However, writers looking to grow would be better served by a Twitter content creator platform and the built-in distribution. Twitter would then hope to retain the creators as they grow and scale, since there would be costs to moving from platform to platform. Because Twitter also offers distribution to its large userbase, they can justify charging more than the barely sustainable 10% fee which Patreon and Substack charge; Facebook, Twitch, and YouTube charge up to 50%.
So what would a Twitter subscription product look like? Some predictions, in order of likelihood:
Subscriber-only longer-form or audio-based content. Twitter has been setting the groundwork here, moving away from their 140 character days with shifts to 280 characters, easier tweet threading, and Moments. The recent audio snippet beta also lines up very closely with shorter podcasts, which there is clearly consumer demand for.
Subscriber-only replies or DM’s. Twitter has been building stronger controls over who can engage with content already (e.g. only other tagged users or people the author follows), and this would be a natural extension. There are a bunch of variations on what this might look like, e.g. how anyone can reply to any tweet but blue checks are guaranteed to show up first, so something analogous for subscribers.
Donations/tipping.
Subscriber-only Periscope chats. This would be super cool, as a way to engage with creators, but it would require Twitter actually doing forward development on one of their acquired products. The horror!
A partnership with news organizations to sell subscriptions to paywalled content accessed via the Twitter app or web interface. My gut is that there is a low probability of happening - high upfront expense, reasonably high chance of some success, but low return given how difficult partnerships are with news organizations and needing to split revenue. The job listings talk about reusability of the subscription code, so I expect they’ll want to do something outside of just the core product.
Will this product get built, will Twitter execute well, will it cannibalize ads focus, etc, who knows! But I think it’s the logical move for Twitter and one which could kneecap the growth of third-party monetization platforms, by effectively solving the biggest creator need: finding readers.