Blog
FoundersiOSProductArchitecture

Most Startups Should Not Build Their Own Subscription Billing

Vishal Paliwal

Vishal Paliwal

Founder & CEO · November 15, 2026 · 7 min read

Most startups should not build their own subscription billing layer. Apple's StoreKit 2 makes the purchase sheet easy, and that is exactly why founders get fooled: the purchase is the smallest part of a subscription business. The expensive part is everything that happens in the months after the first payment.

At Eightinity Technologies we build native iOS apps in SwiftUI, and on any app with a paywall we start from a different question than "which SDK?" We ask who will own the subscription state when a card fails at 2 a.m. For most early-stage products, the answer should be a managed service such as RevenueCat, not your own server.

Why is the purchase sheet the easy part?

StoreKit 2 gives you async Swift APIs, signed transactions, and on-device entitlement checks. A developer can show a paywall and enable a feature in days. A demo looks finished at that point, and that is where the trouble starts.

A live subscription is a state machine. Someone subscribes, renews, upgrades, downgrades, asks for a refund, changes device, lets a card expire, and sometimes comes back months later. Your app has to be right about every one of those states, on every device, every time it opens.

What does Apple actually expect you to handle?

Apple's own subscriptions guidance is a useful reality check. It says that when a renewal fails, Apple attempts to recover it for 60 days. It recommends enabling a billing grace period of 3, 16, or 28 days so a subscriber keeps access while the billing problem is fixed. It also points developers to App Store server notifications, including a failed-renewal notification type, and to a server endpoint that tells you whether a subscription is active, expired, in billing retry, or in grace period.

That list translates into real work:

  • A backend endpoint that receives and verifies signed App Store notifications.
  • Logic that maps each notification to an access decision in your product.
  • A way to keep web, iPhone, and iPad users in the same entitlement state.
  • Handling for refunds, price-increase consent, and plan changes.
  • Monitoring, so a broken webhook does not silently lock out paying users.

None of this is exotic. All of it is unglamorous work that a founder will not notice until it breaks.

When does building it yourself make sense?

There is a real counter-argument, and it deserves a fair hearing. A managed service takes a share of revenue. RevenueCat's account documentation says it is free until monthly tracked revenue reaches $2,500, and then bills 1 percent of revenue on the Pro plan. Check the current terms before you commit, because pricing changes.

If your app grows into serious volume, that percentage becomes a line item a finance lead will question. At that point, building your own server, with a team that can maintain it, can be the right call. We do not argue otherwise. We argue about timing: most founders are nowhere near that scale when they decide, and they pay for the decision in launch weeks.

What does the choice look like side by side?

QuestionManaged serviceStoreKit 2 plus your own server
Time to a working paywallShort. Products and entitlements are configured, not coded.Longer. Server endpoints and state logic come first.
Who handles renewal edge casesThe vendor, with your team configuring rules.Your team, permanently.
Cross-platform entitlementsBuilt in.You design and maintain it.
Ongoing costA percentage of tracked revenue above a free tier.Engineering time and hosting.
Control and data ownershipShared with a vendor.Full.
Best forEarly and growth-stage products.High-volume products with a backend team.

What happened when a login flow was underestimated?

We have a scar that explains our bias. On SOAPNoteAI, a live web product we took to full iOS, a sign-in flow we expected to be quick took many iterations before it worked end to end on a real device. The lesson was not about OIDC. It was that anything involving identity and server state looks small in the plan and behaves large in the build.

Subscription entitlements are the same species of problem. They sit between your app, Apple, and your backend, and the failures only show up under real-world conditions. Every week you spend building that bridge is a week not spent on the product your customers are paying for.

What should be set up before the first paywall ships?

Whichever route you choose, a few decisions are cheap on day one and painful in month six. Name your products and subscription groups carefully, because renaming them later touches live subscribers. Decide what a lapsed subscriber sees, whether that is a locked screen, a read-only mode, or a win-back offer.

Write down your refund and support policy before launch too. Someone on your team will receive a message from a paying user who is locked out, and they need an answer, an owner, and a way to check the user's real subscription status. Teams that skip this treat billing as a feature. Teams that do it treat billing as an operation, which is what it becomes the moment real money moves.

Test the unhappy paths in Apple's sandbox before you test the happy one: an expired card, a cancelled plan, a restored purchase on a second device. These are the cases reviewers and customers hit first, and they are the ones a demo never shows.

How should a founder decide this week?

  1. Count your platforms. If a subscriber can also log in on the web or Android, a shared entitlement layer is worth paying for.
  2. Estimate your first-year revenue. If it is far below the point where a 1 percent fee hurts, the fee is cheaper than the engineering time.
  3. Ask who is on call. If nobody on your team will watch notification failures, do not build the system that produces them.
  4. Keep an exit. Store product IDs and user IDs in your own database so you can move later without migrating subscribers blindly.

Pricing the subscription itself is a separate decision. Before you set it, read how Apple's commission can drop to 15 percent for small developers, and plan your first six months after launch with billing support time included.

FAQ

Should I use RevenueCat or build my own subscription backend?

Most startups should use a managed service such as RevenueCat first. It removes server work for renewals, refunds, and cross-platform access, so you launch sooner. Build your own only when revenue is large enough that the fee outweighs a dedicated team maintaining billing logic, and you have engineers to run it.

Is StoreKit 2 enough on its own for an iOS subscription app?

StoreKit 2 handles purchases and on-device entitlement checks well, which is enough for a simple iPhone-only app. It is not enough once you need web or Android access, analytics, or reliable handling of billing retries and refunds, because those need server-side state that StoreKit alone does not give you.

What happens when a subscriber's card fails on iOS?

Apple attempts to recover a failed renewal for up to 60 days. If you enable a billing grace period of 3, 16, or 28 days in App Store Connect, the subscriber keeps access while Apple retries. Your app or server should read subscription status so access reflects the true state.

How much does RevenueCat cost for a small app?

According to RevenueCat's documentation, the platform is free until an app reaches $2,500 in monthly tracked revenue, and then charges 1 percent of revenue on the Pro plan. Terms change, so confirm current pricing on RevenueCat's site before budgeting for it.

Designing a paywall and not sure what sits behind it? Bring your pricing plan to a free 30-min call and we'll map the billing risks →

Share

Eightinity Technologies

Building something like this?

We'd love to hear about your product. 30 minutes, no pitch — just a direct conversation with the team that ships it.

Book a free strategy call→

Keep reading