The pitch was simple and it landed precisely because it was true in the ways that mattered most at the time. Twitter was collapsing audience relationships every time it changed an algorithm. Facebook had long since stopped reliably delivering content to the people who had asked for it. The newsletter, by contrast, went directly to the reader’s email inbox. No algorithm mediated the delivery. No platform could decide, arbitrarily, to stop showing your work to the people who had said they wanted it. The email list was yours. You owned your audience.
Alison Roman understood this. In the years after leaving her cookbook empire and her magazine column, she built a newsletter on Substack with 343,000 subscribers. Then, in 2025, she moved it to Ghost. Anne Helen Petersen, one of Substack’s most prominent writers and a recipient of one of its development advances, left for Patreon. Lyz Lenz departed citing bot subscribers depressing her engagement metrics and what she described as Substack’s recommendation engine pushing “rage, Nazis, transphobia, and conspiracies.” By March 2025, nearly 3,000 creator migrations away from Substack had been recorded in a single twelve-month period, according to Beehiiv’s own reported figures.
What the departures revealed, and what has become clearer in the period since, is that the original claim — you own your audience on email — was accurate as a relative statement and misleading as an absolute one. Compared to Instagram or Twitter, email gives writers substantially more control. Compared to not being on any platform at all, email is a distribution channel governed by a stack of institutional actors, algorithmic decisions, and terms of service that the writer did not write and cannot override.
The Substack layer
The first layer of platform dependency in the newsletter model is Substack itself, and its terms are more platform-like than the original pitch suggested. Substack’s terms of service govern what can and cannot be published. Its content guidelines restrict promotional and commercial content beyond editorial work. Its fee structure extracts 10% of all paid subscription revenue — at $100,000 a year in subscriber income, that is $10,000 to the platform before a single operating expense is paid, on top of Stripe’s payment processing fees. These are not unusual terms in the media platform economy. They are the terms of a platform relationship, which is what Substack is, whatever the “own your audience” framing implied.
More specifically, Substack has made product decisions that affected audience relationships in ways writers did not control and, in some cases, did not initially know about. The platform introduced a “follow” feature that allows readers to follow a publication without subscribing by email — building a relationship between reader and publication inside Substack’s own ecosystem rather than in the writer’s email list. For writers who understood their subscriber count as representing people who had opted into email delivery, the discovery that a portion of that count represented platform follows rather than email subscriptions was a meaningful recalibration. The list they thought they owned was partially a list of people who had agreed to read inside a platform they didn’t own.
The inbox layer
The more fundamental problem, and the one the original platform-escape narrative did not adequately address, is that the email inbox is not a neutral channel. It is a space governed by Google, Apple, Yahoo, and Microsoft — companies with their own algorithms, policies, and incentives that determine whether any given email reaches any given reader.
Gmail’s spam filters, promotional tab categorization, and engagement-based deliverability algorithms make decisions about every newsletter sent to a Gmail address. These decisions are not arbitrary — they’re based on open rates, engagement signals, spam complaints, and sender reputation — but they are decisions made by Google about what reaches Gmail’s users, and the newsletter writer has limited direct visibility into them and no appeal process when something goes wrong. When a Substack newsletter starts landing in the Promotions tab instead of the Primary inbox, open rates fall, engagement metrics decline, deliverability reputation worsens, and the spiral compounds. The writer did not change anything. Google’s systems made a judgment.
Deliverability is not a solved problem in newsletter publishing — it is an ongoing technical negotiation between the sender’s infrastructure, the recipient’s email provider, and the engagement behavior of the list. Substack manages the infrastructure layer, which is a service it provides reasonably well. But it does so on behalf of thousands of publications simultaneously, which means individual writers have limited ability to customize their sending behavior, authenticate their domain in ways that would improve deliverability for their specific situation, or troubleshoot problems at a granular level. The writer who was told they owned their audience has outsourced the delivery of that audience to a company that has outsourced part of that delivery to Gmail.
Apple’s Mail Privacy Protection, introduced in 2021 and now widely adopted, added another layer to the problem. MPP masks whether an email was opened, making open rates unreliable as engagement metrics and complicating the feedback loops that writers and platforms use to assess list health. A list that appears, based on open rates, to be performing at 30% engagement may be performing at substantially less; the inflation introduced by Apple’s pre-loading behavior makes the signal noisy in ways that affect both the writer’s understanding of their audience and the deliverability reputation signals that email providers use to make routing decisions.
What “owning” actually means
The thing that newsletter writers genuinely own is the data: the email addresses of their subscribers, which can be exported from Substack at any time and moved to another platform. This is a real and meaningful protection. It means that if Substack changes its terms in ways that become unacceptable, or collapses as a business, the writer can migrate their list to a different platform. The audience relationship is not entirely at the mercy of the host platform, in the way that a Twitter following or an Instagram account is. This is the legitimate version of the ownership claim, and it is the reason the newsletter model is structurally more defensible than social media followership.
But what owning the data does not mean is owning the distribution. The email addresses are portable. What those email addresses represent — the behavior, the engagement, the delivery — is governed by systems the writer does not control. Moving a list from Substack to Ghost or Beehiiv or Mailchimp changes which platform’s rules the writer lives under and what fee structure they pay. It does not change the fact that the list still sends to Gmail inboxes filtered by Google’s algorithms, to Apple Mail processed by Apple’s privacy infrastructure, to Yahoo accounts governed by Yahoo’s deliverability policies.
This is, in a sense, the general condition of publishing in a networked environment. There is no truly platform-neutral distribution. Every channel through which writing reaches readers is governed by someone: the editor, the algorithm, the platform, the email provider. The question is never whether you’re on a platform; it is always which platform, whose rules, and what your options are when those rules change.
What the migration wave reflects
The writers who left Substack in 2025 were not, for the most part, leaving because they had found a channel with no platform. They were leaving because specific aspects of Substack’s particular set of rules had become unacceptable to them — the content moderation environment, the fee structure at scale, the limitations on building more complex businesses — and because the friction of moving a portable list was lower than the friction of staying.
That is a different story from the one that launched the newsletter era. The early Substack narrative was about escape from platforms. The 2025 departures are about choosing between platforms — selecting which set of constraints, fees, and distribution dependencies best serves a particular kind of publishing practice.
The email inbox was always a platform. The companies that deliver to it have always made decisions about what arrives. Understanding this doesn’t diminish the value of the newsletter model — it remains a more defensible distribution channel than social media for writers who want to maintain direct audience relationships. What it changes is the framing, and framing matters for the decisions writers make about where to invest in building.
The question is not whether to be on a platform. The question is which platform’s interests are most aligned with yours, and what it costs you when they diverge. Those are questions worth asking before you build 343,000 subscribers on any particular answer.
