Skip to content
AppDues
Log in
Contact salesSign up

Migrating a Custom iOS Paywall to Dashboard Paywalls

A custom paywall can look exactly how you want and still be expensive to change. Every A/B test becomes a code change, a build, and an App Store review. This guide walks through a migration pattern that keeps the old paywall in place while a dashboard-driven paywall is tested beside it, then shows how to read the results without over-claiming what they mean.

Why move a hand-built paywall to a dashboard-driven one

The main trade-off is control versus iteration speed. A fully custom paywall gives you pixel-level design freedom, but each experiment requires new code. A paywall configured from the subscription platform dashboard lets you change layout, product visibility, and promotional variants without shipping an app update.

That does not mean the design has to be thrown away. The FooterView rendering mode lets a dashboard-driven footer sit inside an existing custom paywall, so the surrounding design and animations can stay while the purchase area becomes configurable from the backend.

Before starting, confirm three things:

  • Your products and Offerings are already modeled in the subscription platform.
  • You can identify which Offering your app currently reads at purchase time.
  • You have a way to branch on Offering metadata in your app code.

If any of those is missing, fix it first. The migration pattern below depends on all three.

The migration pattern, step by step

The approach is a parallel Offering rather than a replacement. Nothing about the old paywall is deleted until the new one has proven itself.

  1. Duplicate the Offering that currently drives your custom paywall. In the subscription platform dashboard, open the Offerings tab, use the ellipses under Actions, and choose the duplicate option. Give the copy a clear name such as "Using RC Paywalls".
  1. Add a boolean to the Metadata of both Offerings. In the duplicated Offering, add a key named rc-paywall and set it to true. In the original default Offering, add the same key and set it to false.
  1. Branch in code on that metadata value. When the app receives an Offering with rc-paywall set to false, render the existing custom paywall. When it is true, render the dashboard-configured paywall.
  1. Ship the update and let it reach most users before starting any experiment. The experiment only works if the code that reads the metadata is already live on the devices you are measuring.

The two Offerings are otherwise identical, so the only variable is which paywall renders. The exact code that performs the branch is supplied as a protected module and should be used as-is rather than retyped from this description.

Field and control map

Use this map to keep the dashboard configuration and the app code in sync.

  • Offering: the container the app reads at purchase time. Two are needed during migration: the original and the duplicate.
  • Metadata: per-Offering key/value store. This is where the switch lives.
  • rc-paywall (boolean): true means render the dashboard-configured paywall; false means render the custom one.
  • condensed property: on the footer view, controls whether one product or all products show by default.
  • Experiments tab: where two Offerings are compared. The + New button starts a new experiment.
  • New customers to enroll slider: the percentage of new users entering the experiment. Half of those see the treatment.

A common mistake is to put the switch in app code as a hardcoded flag. That defeats the purpose, because changing it would again require a release. Keep the switch in Offering metadata so it can be flipped from the dashboard.

Worked example: the enrollment math

The enrollment slider is easy to misread. It controls how many new customers enter the experiment, not how many see the new paywall.

If the slider is set to 20%, then 20% of new users are enrolled. Within that enrolled group, the traffic is split between control and treatment, so about half see the dashboard-configured paywall. The effective share of all new users seeing the new paywall is therefore roughly 10%.

A cautious sequence:

  • Day one: enroll a small share, such as 20%, and confirm that purchases complete correctly on the new paywall.
  • After confirmation: raise enrollment to 100%, which means about 50% of users see each paywall.
  • During the run: check daily for a catastrophic drop, but otherwise let enough customers accumulate to make the comparison meaningful.

This staged approach is the difference between a controlled migration and a live gamble on your revenue.

Reading the results without over-claiming

The first experiment in the original walkthrough compared the old custom paywall with the new dashboard-configured paywall built to look almost the same. The author reported that, after two weeks, initial conversion rate was up 4.76% across roughly 11,000 customers, and he stated plainly that this figure was not necessarily statistically significant. Those are his captured, dated results from that walkthrough, not an independent benchmark.

That framing matters. The purpose of that first experiment was not to find a winner but to confirm that nothing broke. A near-identical design producing a near-identical conversion rate is the desired outcome.

When reading any paywall experiment, separate three questions:

  • Did anything break? Check purchase completion and error rates first.
  • Did the metric move? Compare initial conversion, paid conversion, and revenue, not just one of them.
  • Is the movement real? Consider sample size and whether the difference could be noise.

A lift that is not statistically significant is not proof of improvement. It is a signal that may justify a longer run or a larger sample.

Experiment example: one product versus all products

A second experiment tested a layout change rather than a migration. The control showed only the Annual plan by default, with Monthly and Lifetime behind an "All Plans" button. The treatment showed all three products immediately.

The author's captured, dated results show why a single headline metric can mislead. Monthly first-time conversions rose about 242% and Lifetime about 266%, while Annual fell roughly 30%. Even after those shifts offset each other, overall initial conversions still rose by nearly 10%. Churn climbed sharply, by about 57%, which the author interpreted as an expected effect of monthly subscribers now forming a larger share of the customer mix.

Read together, these numbers tell a more complicated story than the headline conversion lift. More people started a subscription, but a larger portion of them chose the shortest commitment. Whether that is good depends on realized lifetime value over time, not on the first conversion alone.

{{OWNER_CONFIRM_400_DAY_MEASUREMENT}} The author noted that the treatment's realized LTV was still rising about a month after stopping, and that a full year would be needed to compare annual renewals across both cohorts.

Trial length and time-limited promotions

Trial length is another variable that can be tested from the dashboard. In the original walkthrough, a 3-day trial was compared with a 7-day trial, and early results favored the longer trial. Those results were described as preliminary, so treat them as a direction to investigate rather than a settled conclusion.

Promotions follow the same pattern. A Black Friday variant was built as a separate paywall in the dashboard and pushed out by making that Offering current, without an app update. The author also pointed to scheduled targeting as a way to automate that switch: {{OWNER_CONFIRM_SCHEDULED_TARGETING}}

If your app can render either a footer paywall or a full-screen dashboard-configured paywall based on Offering metadata, a seasonal promotion becomes a configuration change rather than a release. That flexibility is the practical payoff of the migration.

Pre-flight checklist

Before you start, confirm each item:

  • The current Offering is identified and can be duplicated.
  • Both Offerings carry the rc-paywall metadata key with opposite boolean values.
  • The app reads that key and branches to the correct paywall.
  • The update has propagated to most users before experiments begin.
  • A first experiment compares old versus new at low enrollment to catch breakage.
  • Enrollment is raised only after purchases are confirmed working.
  • Results are read across initial conversion, paid conversion, revenue, churn, and LTV.
  • Sample size is checked before treating any lift as real.

If any item is unchecked, pause. The cost of a broken paywall is measured in lost revenue, not in extra engineering time.

Frequently asked questions

Do I have to rebuild my paywall design from scratch?

No. The FooterView rendering mode lets a dashboard-driven footer sit inside an existing custom paywall, so the surrounding design can be preserved while the purchase area becomes dashboard-configurable.

What does the rc-paywall metadata key do?

It is a boolean stored on an Offering. When set to true, the app renders the dashboard-configured paywall; when false, it renders the custom paywall. Because it lives in Offering metadata, it can be changed from the dashboard without a new app release.

If I set enrollment to 20%, what share of users sees the new paywall?

About 10%. The slider controls how many new customers enter the experiment, and those enrolled users are split between control and treatment. Half of 20% is roughly 10% of all new users.

How long does the platform keep measuring an experiment after it stops?

{{OWNER_CONFIRM_400_DAY_MEASUREMENT}}

A conversion lift looked good but churn rose. Is that a win?

Not necessarily. A shift toward shorter commitments can raise initial conversions while increasing churn. The real answer depends on realized LTV across cohorts over time, not on the first conversion metric alone.