Substack Keeps 10%, but a $5 Sub Loses 20% Before Apple Touches It

A $5/month Substack subscriber nets the writer $4.01, and an iOS sign-up in year one stacks Apple's 30% on top of that.

---

Charge $5 a month on Substack and the writer keeps $4.01. Substack's headline fee is 10%, so the round number a creator carries in their head is $4.50. The $0.49 gap between $4.50 and $4.01 is Stripe, and it is the first sign that the advertised platform fee covers a fraction of what a subscription costs to run.

beehiiv published the full breakdown on a $5 charge: $5.00 in, minus Substack's 10% ($0.50), minus Stripe's 2.9% ($0.15), minus Stripe's flat $0.30 per transaction, minus the 0.7% recurring-billing fee ($0.04), leaving $4.01. Total fees run roughly 20%, double the headline, because Stripe's flat $0.30 is a 6% charge on a $5 sub and shrinks to a rounding error on a $50 one. Substack's 10% holds at every price. The processing fee climbs as the price falls, and most creators charge at the low end where it bites hardest.

For a creator-CEO pricing a newsletter, a community, or a paid feed, the operating question is not which platform advertises the lowest fee. It is what the full stack takes once payment processing, the app-store tax, and the subscribers lost to failed payments are added to the platform's cut. Three layers sit on top of the headline number, they compound, and they hit different creators by different amounts depending on price point and where the subscriber signs up. This guide builds the real take from published fee schedules, plus one clearly labeled churn range, so a creator can run their own.

Layer one: Stripe, the fee that scales backwards

Substack runs subscriptions on Stripe, and so do most of its rivals. Substack's own support page states that writers keep "90% of their revenue minus credit card fees", which concedes in one clause that the 10% is not the whole cost. Those credit-card fees are Stripe's: a standard US online rate of 2.9% + $0.30 per successful charge, plus a recurring-billing fee that Stripe lists at 0.7% of billing volume on its pricing page. That 0.7% rose from 0.5% on July 10, 2024, so every subscription created since pays the higher stacked rate.

The percentages hold across price points. The flat $0.30 does not. On a $5 charge the $0.30 alone is 6% of gross, on a $10 charge it is 3%, on a $50 charge 0.6%. Add Substack's 10% and Stripe's 2.9% + 0.7% to that fixed cost and the all-in fee on a $5 sub clears roughly 20%, while the same stack on a $50 sub runs closer to 14%. The headline fee describes the percentage layer accurately and leaves out the fixed-cost layer, which is the one that decides whether a $5 price is viable.

The clean baseline for comparison is Ghost, which charges a 0% transaction fee and connects creators directly to their own Stripe account, leaving Stripe's 2.9% + 30¢ as the only cost. Measure every other platform against that floor: the gap between a platform's all-in take and Ghost's Stripe-only floor is what the platform charges for hosting, distribution, and discovery.

Layer two: the app-store tax, which dwarfs the rest

The largest single deduction belongs to Apple. Apple's developer terms state that "during a subscriber's first year of service, you receive 70% of the subscription price," a 30% cut, rising to 85% only after a subscriber completes a full paid year. A subscription bought inside an iOS app in year one therefore stacks a 30% commission on top of the platform-and-processing layer underneath it.

Substack handles this by raising the in-app price. Its IAP FAQ confirms the mechanics: Substack continues to take its 10% on the original web price, automatically sets a higher price in the iOS app to offset Apple's fee, and does not raise its own share to capture any of Apple's cut. The creator nets approximately the same, and the reader pays the markup. The markup runs higher than 30% because Substack's 10% and Stripe's fees scale up with the inflated price. Isabelle Roughol documented a live case: Nicolas Colin's Drift Signal, priced at £17/month on the web, costs £23 in the Substack app, a 35% jump, because the higher base price drags Substack's cut and Stripe's fees up alongside Apple's 30%. The creator's net holds and the reader absorbs a 35% premium. Whether that pricier in-app screen converts worse is the open question the published data does not answer [NEEDS REPORTING: conversion-rate impact of the inflated iOS price — the implied drop when a reader sees £23 vs £17 — needs creator-reported A/B or platform data].

The 30% is the first-year rate that catches the platform aggregators. An individual developer who bills under $1M a year qualifies for Apple's Small Business Program and pays 15% from day one, but a writer publishing through Substack sits inside Substack's developer account rather than their own, so the individual writer does not get the 15% rate. Google Play is cheaper and flatter: its published service fee is 15% on auto-renewing subscriptions regardless of revenue. The platform a reader signs up on, and the device they use, changes the creator's economics before anyone looks at the platform fee.

Layer three: involuntary churn, the leak you don't see

The third layer is the softest, and it must be labeled as such: it is the subscribers lost to a failed payment — an expired card, a bank decline, a hard expiry — rather than to a cancel button. Recurly's benchmark research puts involuntary churn at 1.38% of subscribers per period across all businesses, rising to 2.16% for B2C, the segment most creator subscriptions sit in. *(Modeled input: this is an industry benchmark, not a measured figure for any specific creator. A creator's real number depends on list age, geography, and whether they run payment retries.)*

The leak compounds because it is unrecovered revenue, not a one-time deduction. Butter Payments estimates the average subscription business loses about 10% of annual recurring revenue to failed payments each year, and Churnkey, citing industry data, puts involuntary churn at 20–40% of total churn — revenue that disappears unless dunning (automated retries on failed cards) claws it back. Layer that on top of a low price, a heavy iOS mix, and an un-retried list, and the share a creator actually banks slides past a 25% effective cut of what subscribers nominally paid. That figure is the modeled top of the stack, not a published rate, and it is why the headline number understates the real cost most for the creators who can least afford it: cheap subscriptions, mobile audiences, no dunning.

The published fee schedules, corrected

A common framing puts Skool and Whop at "around 9%." Their published schedules do not say that. Skool charges 10% + $0.30 per transaction on its Hobby plan and 2.9% + $0.30 on Pro (rising to 3.9% above $900). Whop's documented base is 2.7% + $0.30 per domestic transaction, with +1.5% for international cards, +1% for currency conversion, an optional 0.5% billing fee, and payout fees such as $2.50 for next-day ACH. Patreon moved all new creators to a flat 10% platform fee after August 4, 2025, replacing its legacy 5/8/12% tiers; creators with a page published before that date keep the old rates only while the page stays published. Gumroad charges 10% + $0.50 for direct sales [NEEDS REPORTING: Gumroad's processing-fee line and 30% Discover-marketplace rate to be confirmed against the live pricing page]. Ghost stays at 0% plus Stripe.

Read these the way the stack reads them: the platform fee is one input, the processing line is a second, the app-store cut applies only to in-app sign-ups, and churn applies to everyone. Two creators on the same platform at the same advertised fee can keep materially different shares depending on price and channel.

What to compute before you price

Run your own number in four steps. Start with the platform fee from the published schedule. Add Stripe's 2.9% + 0.7% and the flat $0.30, then check what that $0.30 is as a percentage of your price — on anything under $10 it is the line that moves your economics. Add Apple's 30% (year one) only against the slice of subscribers who sign up inside the iOS app. Then subtract a churn allowance — use the 2.16% B2C benchmark as a starting estimate, adjusted for whether you run retries.

The output reorders the pricing decision. A $5 iOS-heavy newsletter with no dunning can hand back more than a quarter of gross before the writer sees a cent; the same content at $10 on the web, with retries on, keeps far more of every dollar. Platforms compete on the advertised 10%, but the number a creator should price against sits at the bottom of the stack, after Stripe, Apple, and churn have each taken their cut.