Here is the number that settles most of the ghost vs substack argument before it starts. A newsletter with 10,000 members and 500 readers paying $8 a month grosses $48,000 a year. Substack’s help center says the platform keeps 10% of subscription revenue, so that writer hands over $4,800 a year before Stripe touches a cent. On Ghost(Pro), the Publisher plan for a list of up to 10,000 members costs $1,056 a year, according to the pricing data Ghost publishes on its own pricing page, and Ghost takes 0% of paid memberships. Same readers, same price, roughly $3,700 a year apart.
That gap is real, but it is only one of three costs you are choosing between. The other two are harder to see on a pricing page: what it costs you to leave, and what the platform’s audience network is worth to you. This piece walks through all three, with current numbers, and gives you a test you can run on your own newsletter in about fifteen minutes.
The fee math at 500 paid readers
Start with what both platforms share. Substack and Ghost both run payments through Stripe, and both connect to a Stripe account in your name. Substack’s support page lists Stripe’s US card fee as 2.9% plus 30 cents per payment, plus a 0.7% Stripe Billing fee for recurring charges. Ghost’s documentation names the same 2.9% plus 30 cents. Stripe Billing applies to recurring subscriptions wherever Stripe runs them, so check your own Stripe dashboard, but assume parity for planning.
Run the 500-reader example through it. At $8 a month, 500 readers produce 6,000 card charges a year. The 2.9% cut on $48,000 is $1,392. The 30-cent fixed fee on 6,000 charges is $1,800. Stripe Billing at 0.7% adds $336. Total Stripe cost: $3,528 a year on either platform.
Now add the platform layer. On Substack, 10% of $48,000 is $4,800, so total fees come to $8,328 and the writer keeps $39,672. On Ghost(Pro) Publisher at the 10,000-member tier, the hosting bill is $1,056 a year, total fees come to $4,584, and the writer keeps $43,416. The difference is $3,744 a year, or about 7.8% of gross revenue.
Two details change the math more than people expect. First, Ghost’s cheapest plan cannot sell subscriptions. Starter costs $15 a month billed yearly ($18 billed monthly) for up to 1,000 members, and Ghost’s feature table reserves paid subscriptions for Publisher and above. Second, Ghost prices by member count, including free members. A writer with 100,000 free readers and 300 paid readers pays Ghost $3,288 a year on Publisher, which may exceed what Substack’s 10% would cost on that small paid base. Substack’s fee scales with revenue. Ghost’s fee scales with audience size. Which one hurts depends on your conversion rate, not your list size alone.
Ghost’s prices also moved recently. TechCrunch reported in August 2025, alongside the Ghost 6.0 release, that entry hosted plans rose from $9 to $15 a month and from $25 to $29 a month, with reduced pricing for publishers with large subscriber counts. If you priced Ghost two years ago, price it again.
Where is the break-even line for your list?
The cleaner way to settle ghost vs substack on cost is to find the paid-subscriber count where Substack’s 10% equals Ghost’s flat fee. Below that line, Substack is cheaper. Above it, Ghost is cheaper, and the gap widens with every new paid reader.
The formula fits on a sticky note. Divide Ghost’s annual plan price for your member tier by 10% of one subscriber’s annual payment. At $8 a month, one subscriber pays $96 a year, and Substack keeps $9.60 of it.
For a list of up to 1,000 members, Ghost Publisher costs $348 a year. $348 divided by $9.60 is 36.25, so the break-even sits at 37 paid readers. For a list of up to 10,000 members, Publisher costs $1,056 a year, and the break-even is 110 paid readers, or 1.1% of the list. For a list of up to 100,000 members, Publisher costs $3,288 a year, and the break-even is 343 paid readers.
Those three numbers tell you something useful. Most newsletter operators who charge at all convert more than 1.1% of their list to paid. If you are at 10,000 members and charging $8, you probably crossed the Ghost line a while ago. If you are a large free list with a thin paid tier, Substack’s revenue-based fee may still be the cheaper bill.
Lower prices push the line up. At $5 a month, Substack keeps $6 per subscriber per year, and the 10,000-member break-even rises to 176 paid readers. Annual plans push the line around too, because fewer charges mean fewer 30-cent fixed fees, which helps both platforms equally but does not touch the 10% platform cut.
Run your own numbers with your real price, your real member count, and Ghost’s current tier for that count. Do not use someone else’s example, including this one.
Who owns the reader, the card, and the domain?
Fees are what you pay to stay. Control is what you pay to leave, and it comes in three parts.
The reader relationship is the first. Substack’s help center states that you own the email addresses of readers who subscribe through the Substack app, that they appear in your Subscribers tab, and that you can export the list at any time. Ghost is open-source software, and on Ghost(Pro) your member list lives in a Ghost install you can export or self-host. On paper, both let you take your list.
The card relationship is the second, and it is where the details matter. Because both platforms run through a Stripe account in your name, the billing relationship with each paying reader sits in your Stripe account, not in the platform’s. Ghost’s migration guide tells Substack writers to connect “the same Stripe account that is connected to your Substack” and then upload the paid subscriber CSV. That is why paid moves are possible at all. The same guide flags two catches. Substack’s 10% fee continues on existing paid subscriptions after the move until it is removed, and Ghost’s concierge team helps with that. And your readers’ bank statements may still say “Substack” until you update the descriptor in Stripe’s public details.
The domain is the third. Substack offers a custom domain for a one-time $50, according to its pricing help page. Ghost(Pro) includes custom domains on every plan. If you publish on yourname.substack.com, every link, citation, and search result points at Substack’s domain. When you leave, those links break or redirect at Substack’s discretion. If you publish on your own domain from day one, on either platform, the address readers and search engines remember stays yours.
That last point matters more each year, because the place people discover writers keeps shifting. Google, ChatGPT, Perplexity, and Gemini all cite web pages, and a citation pointing at your domain keeps paying you after you change software. A citation pointing at a platform subdomain pays you only while you stay. If you want to see how often answer engines currently surface your work, Instant Press runs a free AEO rating that checks exactly that.
What Substack’s network is worth in subscribers
The strongest case for Substack is not the editor or the price. It is the network. Substack’s own team wrote in a February 22, 2024 post that its Recommendations feature and app together drive 50% of all new subscriptions and 25% of new paid subscriptions on the platform. An earlier Substack post put the figures at more than 30% of new free subscriptions and around 10% of paid subscriptions, so the network’s share grew as the app grew.
Treat those as platform-wide averages, not your forecast. A writer who already has a large audience on X, LinkedIn, YouTube, or a podcast will see a smaller share from Substack’s network, because their own channels do more of the work. A new writer with no audience may see a larger share, because the network is the only channel they have.
Ghost has been building its own distribution answer. Ghost 6.0, released August 5, 2025, connected Ghost publications to the open social web through ActivityPub, so readers on Mastodon, Threads, Flipboard and other ActivityPub apps can follow a Ghost publication directly, according to TechCrunch. Bluesky works through a Bridgy Fed integration rather than native support. The same report said publishers on Ghost have generated over $100 million. That is a meaningful network, but it is not the same thing as a recommendation engine inside an app that millions of readers open daily. Be clear-eyed about which one you would get.
The useful question is how many of your paid readers the network produced last quarter, not whether the network is good. Substack’s dashboard shows where subscribers came from. If the network share of your paid growth is small, you are paying 10% for something you are not using.
Run the Toll, Door, and Crowd Test
Here is the framework. It scores the three costs that decide this choice, and it takes about fifteen minutes with your Substack or Ghost dashboard open. Call it the Toll, Door, and Crowd Test.
The Toll is your annual platform fee in dollars, ignoring Stripe since both platforms pay it. On Substack, that is 10% of last year’s paid revenue. On Ghost, that is the yearly price of the Publisher tier for your member count. Write both numbers down and subtract. Score the Toll from 0 to 3. A gap under $500 a year scores 0. $500 to $2,000 scores 1. $2,000 to $5,000 scores 2. Over $5,000 scores 3. A higher Toll score favors Ghost.
The Door is how hard it would be to leave your current platform in a bad month. Ask three questions and give one point for each “no.” Do you publish on your own custom domain? Have you exported your full subscriber list in the last 90 days? Do you know which Stripe account holds your paid subscriptions and that you can log in to it? A Door score of 0 means you can walk out any time. A Door score of 3 means you are renting a house with no key. A high Door score does not tell you to switch. It tells you to fix the door first, wherever you stay.
The Crowd is the share of last quarter’s new paid subscribers who came from the platform’s own network. On Substack, pull it from the subscriber source data. Score it 0 to 3 in reverse. If the network produced more than 40% of your new paid readers, score 0. Between 20% and 40%, score 1. Between 5% and 20%, score 2. Under 5%, score 3. A higher Crowd score means the network matters less to you, which favors Ghost.
Add the Toll and Crowd scores, which together range from 0 to 6. A combined 0 to 2 says stay on Substack, because the network is earning its fee. A combined 3 or 4 says you are in the zone where the decision turns on taste, tools, and how much you value the Substack app. A combined 5 or 6 says the 10% is a tax you are paying for a service you mostly do not use. Then look at the Door score separately. Whatever the combined score says, a Door of 2 or 3 is your first job this week.
The test works in reverse, too. A Ghost writer who scores a low Toll gap and has no meaningful social following might find that Substack’s network is worth trying, especially early.
What Platformer’s 2024 move teaches
The best public case study is Platformer, Casey Newton’s tech newsletter. On January 11, 2024, Newton published “Why Platformer is leaving Substack.” His main reason was content moderation, specifically Substack’s December 2023 position that it would remove accounts only for credible threats of violence and would not commit to proactively removing pro-Nazi publications. He also criticized Substack’s recommendation features, such as Notes and the weekly digest, for promoting publications across the network.
Platformer moved to Ghost. At the time, Newton wrote that the newsletter had grown from about 24,000 free subscribers in its early days to more than 170,000, and that paid subscribers had “grown into the thousands.” He told readers that free and paid accounts would be ported over automatically, and that no longer sharing 10% of revenue with Substack would save the business tens of thousands of dollars a year.
Three lessons carry over to any operator. The money was a bonus, not the trigger. Newton left over policy and the behavior of the recommendation engine, which is the Crowd variable cutting the other way: a network that can push your work to new readers can also push work you do not want associated with you. The move was technically possible because the paid relationships ran through Stripe, which is the Door variable doing its job. And the savings figure fits the Toll math above, since at thousands of paid readers on a tech newsletter, 10% runs well into five figures.
The broader point is that platform risk is not hypothetical. Policies change, fee structures change, and features you did not ask for appear in the product. Your exit plan should exist before you need it.
Pick the platform that fits your next 24 months
The ghost vs substack decision comes down to which cost you would rather carry for the next two years. Choose Substack if you are starting from zero, you have no other channel feeding your list, and the app’s recommendation network is producing a real share of your paid readers. Put it on a custom domain from day one, export your list every quarter, and know your Stripe login. The 10% buys you distribution, and early on that can be worth far more than $1,000 a year.
Choose Ghost if your paid base has crossed the break-even line for your member tier, most of your growth comes from channels you already own, and you want your publication to live on your domain with your design and no platform cut. Budget for the right tier, because Starter cannot take payments and the price rises with free members as well as paid ones.
The short version fits in three lines. Substack’s 10% beats Ghost’s flat fee only below the break-even line, which sits at 110 paid readers for a 10,000-member list at $8 a month. Both platforms keep your paying readers in your Stripe account, so leaving is possible, but only if you own your domain and your exports. And Substack’s network, which drove 25% of new paid subscriptions platform-wide in early 2024, is the one thing worth paying the fee for, so measure your share before you decide.