Since May 2025, apps on the US App Store storefront can send users to a web checkout with no Apple commission and no application process. Apple updated guideline 3.1.1(a) to state that entitlements are not required for US storefront apps to include buttons, external links, or other calls to action. Most guides still describe the regime that existed before that change.
The zero percent is real. It is also provisional in a specific way. An appeals court has already ruled that Apple is entitled to charge something for external links, and a district court has been told to decide how much. Meanwhile every cost Apple used to absorb, sales tax across 45 taxing jurisdictions, chargebacks, dunning, subscription state, moves onto you the day you leave in app purchase.
Here is what the US path actually requires, what it costs, and how to build it so a future fee is a configuration change rather than a rebuild.
Do you need an entitlement to link out in the US?
No. This is the single most common error in guides published on this topic.
Before May 2025, linking out required Apple's StoreKit External Purchase Link Entitlement: a formal application, an Account Holder addendum, Apple review, one link per app, a mandatory system disclosure sheet, and transaction reporting through the External Purchase Server API. That apparatus is the exact thing Judge Yvonne Gonzalez Rogers found to be willful non compliance with her 2021 injunction. On April 30, 2025 she enjoined it for the US storefront, and Apple updated the guidelines the next day.
What that means concretely for a US app today:
You do not apply for anything. You do not need the StoreKit external purchase APIs. You are not limited to one link, so buttons, multiple calls to action, and different destinations are all permitted. You are not bound by the entitlement's reporting duty, because that duty belongs to the entitlement. And the full screen warning describing users as leaving Apple's payment protection is gone, because the court enjoined it and permitted only a neutral redirect notice.
The entitlement still exists as an optional route, and some developers adopt it deliberately. The reasoning is defensive: if the district court reinstates a commission, developers already on Apple's APIs will have the reporting plumbing in place. That is a real argument. It is a choice, not a requirement, and treating it as a requirement means building infrastructure the court struck down.
Enforcement in review is uneven, so budget for a rejection that has to be appealed on the guideline text.
Where the case actually stands, as of August 2026
The 0% is a court remedy in active litigation, and the appeals court has already said it will not stay at zero.
The sequence matters, because most published timelines get it wrong. April 30, 2025: Rogers finds Apple in willful contempt and bars any commission on external link purchases. December 11, 2025: the Ninth Circuit affirms the contempt finding in full, but holds the total commission ban is not an appropriately cabined civil sanction, vacates it, and remands for the district court to set a permissible rate. April 6, 2026: the Ninth Circuit stays its own mandate so Apple can prepare a certiorari petition. April 28, 2026: it reverses that stay, finding Apple had not shown a reasonable probability of certiorari or irreparable harm. May 6, 2026: the Supreme Court declines to pause the contempt order. June 30, 2026: the Supreme Court grants certiorari on the narrow question of whether Apple was properly held in civil contempt, with argument in the term beginning October 2026.
The remand is currently stalled. Apple asked Rogers in early July to pause the rate proceedings until the Supreme Court rules. Epic opposed, arguing a rate has to be set regardless of the outcome and that a Supreme Court decision may not land until as late as June 2027. If the pause is denied, Apple must file its commission proposal within 24 hours. Until that proceeding concludes, the US external commission stays at zero.
So the practical answer to when this resolves is not before the end of 2026, and possibly not before mid 2027. Anyone publishing a firmer date is guessing.
What a reinstated commission would realistically look like
This is where most analysis goes wrong by reasoning from Apple's EU fees.
The Ninth Circuit did not leave the number open. It set a standard: Apple may charge a commission based on the costs genuinely and reasonably necessary for coordinating external links, and no more. Some compensation for intellectual property directly used in completing linked out purchases counts, but the court noted that IP is already used to facilitate in app purchases, so the allowance is limited. Security and privacy features, the justification Apple gave for its 27%, were explicitly excluded from the cost base.
That is a cost recovery standard, not a value capture one. Apple's EU pricing, where store services and acquisition fees stack into double digits, is built on the opposite logic and is a poor guide to what a US court will approve. Any range anchored to EU rates should be treated as a ceiling that the legal standard does not support.
The planning implication is narrower than general uncertainty. Zero has already been called overbroad, so a small nonzero number is the base case. Build so that inserting a percentage fee, applied prospectively from a ruling date, is a configuration entry. Do not hardcode zero into pricing logic, revenue recognition, or margin models.
The infrastructure Apple was doing for you
Apple's commission bought a full seller of record stack. Leaving in app purchase means rebuilding it or renting it.
Payment processing. Stripe's standard US rate is 3.4% plus 30 cents per successful online card transaction. That is the base. Subscription businesses typically add Stripe Billing at 0.7% of billing volume and Stripe Tax at 0.5%, which pushes the realistic all in past 4% before disputes. International cards add 1.5%, and currency conversion another 1%.
Sales tax. Calculation, registration, filing, and remittance in every state where you have nexus. Covered in detail below.
Subscription mechanics. Recurring billing, failed payment retries, dunning, proration, cancellation and reactivation flows. Stripe Billing or Recurly cover most of it. Building it in house is a multi month project that keeps consuming engineering time after launch.
Disputes. Apple absorbs chargeback liability on IAP. On web billing it lands on your processor account, at 15 dollars per dispute regardless of outcome, plus the lost revenue. Elevated dispute rates trigger processor account reviews.
A merchant of record collapses those layers into one contract. The MoR becomes the legal seller, which is what actually moves the tax and dispute liability off you rather than just outsourcing the work. tiun operates this way for SaaS and AI companies, pairing the MoR position with a unified backend covering auth, payments, customer database, and analytics, at 3.4% plus 30 cents with no platform fee, no international card surcharge, and no FX margin. That fee replaces the seller of record stack. It is a separate line from whatever commission Apple ends up permitted to charge, and the two should be modelled separately rather than netted against each other.
Building the stack yourself is entirely valid. The point is that it has a real cost that belongs in the comparison before you decide.
US sales tax is the part that quietly gets expensive
Apple calculates, collects, and remits US sales tax as seller of record. On web billing, that obligation is yours.
There is no federal sales tax. States set their own rates, and county, city, and transit district rates stack on top, so effective rates can differ between neighbouring ZIP codes in the same metro. Chicago layers city, county, and regional transit components onto the state rate, and applies a separate lease transaction tax to SaaS.
Economic nexus, established by the 2018 South Dakota v. Wayfair ruling, means you owe tax in a state once your sales there cross a threshold, with no physical presence required. The thresholds are not uniform and they are moving. Most states sit at 100,000 dollars in annual in state sales. California, Texas, and New York are at 500,000. Many states originally paired the revenue test with a 200 transaction trigger, which caught low priced subscriptions long before they earned six figures in a state, but that test is being repealed: as of 2026 most states have gone revenue only, and Kentucky drops its transaction threshold in August 2026.
Then there is whether your product is taxable at all. SaaS is taxable in roughly 22 to 25 jurisdictions as of 2026, not the 30 figure that circulates, which describes digital products broadly. Pennsylvania taxes SaaS at 6% as prewritten software and its nexus threshold is 100,000 dollars with no transaction prong, which makes it a cleaner worked example than California. California exempts most SaaS entirely on the reasoning that no tangible property changes hands. Texas taxes 80% of the charge under its data processing rule. Washington taxes it as a digital automated service.
Two practical points. Overcollecting in an exempt state is a customer trust problem and a refund liability. Undercollecting in a taxable state is a balance sheet problem, because the state assesses you for tax you never charged, plus penalties and interest. Registration obligations generally begin once you cross a threshold rather than reaching backwards through the year, but the lookback on unregistered periods is where assessments actually come from.
Automate this from launch. Retroactive cleanup costs multiples of doing it correctly the first time.
The conversion problem on the web leg
Some drop off between app and web checkout is guaranteed. How much depends almost entirely on how you build the handoff.
IAP converts because payment credentials are already on file and confirmation is a biometric tap. Your web checkout starts cold. The closest available equivalent is Apple Pay on the web, which restores the saved card and Face ID confirmation. Implement it before anything else.
Mobile layout is not optional, since every user arriving from your app lands on a phone browser. Desktop style form filling and multi page confirmation flows lose people who would have converted through IAP.
Build the return path deliberately. After payment, the user needs to land back in the app with entitlements visibly active. A confirmation page with no route home generates support tickets and disputes from people who believe the charge failed. Test the full round trip, including entitlement verification on return, before launch.
Show the price differential in the app. A specific comparison against your IAP price is the thing that earns the tap. A generic prompt to continue on the website, with no price anchor, performs materially worse. What you cannot do is disparage Apple or in app purchase in that surface.
What to build now
Skip the entitlement unless you have a specific reason to adopt it. Implement Apple Pay on the web and a mobile first checkout. Solve sales tax with automation or an MoR before your first cross state sale rather than after. Build the deep link return and verify entitlements on the way back in. Keep IAP live or maintainable, because a single legal ruling or guideline change can close the external path faster than you can rebuild it. Collect email addresses from web purchasers, so you have a channel that does not depend on in app messaging rules. And treat the commission as a variable, with a zero value today and an editable one tomorrow.
The window is open, the economics are better than IAP for most subscription businesses at volume, and the terms are going to change. Build for the version where they have.
Frequently asked questions
Can I use external payment links while the case is still on appeal?
Yes. Nothing about the Supreme Court review or the pending remand suspends the current rule. Guideline 3.1.1(a) permits US storefront apps to include buttons, links, and other calls to action with no entitlement, and the Supreme Court declined to pause the contempt order in May 2026. What is being decided is the future commission rate, not whether the right to link out exists. You do not need approval from anyone to start.
If a commission comes back, will it apply to transactions I already made?
Almost certainly not. Courts apply fee determinations prospectively as a rule, and retroactive collection on completed transactions would create practical problems the district court has no reason to invite. Plan for a rate that begins on a ruling date and runs forward. What that requires of your systems is the ability to apply a percentage from a specific date, which is a billing configuration question rather than an architectural one.
Can I charge less on the web than I charge in the app?
Yes. You set your web price independently, and displaying both prices inside the app is permitted on the US storefront. Showing 8 dollars per month on your site next to 12.99 through in app purchase is compliant, and that visible gap is usually what drives the tap. The limits are on conduct rather than price: you cannot disparage Apple or in app purchase in that surface, and the link and its context still fall under general App Store guideline terms.
What happens to my existing IAP subscribers if I move to web billing?
They stay on IAP. A subscription is bound to the channel that created it, so there is no mechanism to migrate an active IAP subscriber to your own billing mid term. The workable path is parallel operation: existing subscribers renew through Apple until they churn or lapse, new users go to web checkout, and you decide separately whether to offer an incentive to switch at renewal. Depending on your retention curve, expect to run both systems for 12 to 24 months and budget accordingly.