Define what a conversion means
A paywall view, a trial start and a paid subscription are different events. Track them separately. A useful paywall report begins with a specific question: which eligible viewers started a new subscription, and what happened to their trials afterward?
In Appbase, paywall conversion combines a verified new trial or a verified direct purchase, divided by eligible unique viewers. It does not mean every conversion has paid. Existing subscribers can see a paywall too; retain that behavior while excluding it from new-subscription conversion. Unknown access state should remain visible rather than silently being treated as confirmed inactive.
Separate design from placement and presentation
| Concept | Example | Purpose |
|---|---|---|
| Paywall design / version | Main paywall, version 2 | Compare the experience you shipped |
| Placement | Onboarding or feature gate | Understand the context that brought people there |
| Presentation | One actual showing | Connect a purchase attempt to the right exposure |
The same design can behave differently in onboarding and settings. Repeated presentations are also not additional unique viewers. Keep those distinctions when comparing rates. An onboarding funnel counted by attempt can legitimately differ from a paywall report counted by unique viewer and view date.
Connect behavior to billing evidence
Record the actual presentation, purchase intent, product and transaction identifier through the mobile SDK. Connect RevenueCat on the backend and keep its customer identity aligned with your app’s analytics identity. Appbase uses billing evidence and transaction matching to verify outcomes.
A client success callback alone is not revenue proof. Restores, existing entitlements, duplicate deliveries and interrupted connections can otherwise inflate results. Keep provider secrets on the backend; mobile collection keys are public, write-only credentials.
Follow the RevenueCat integration guide and the paywall report reference. Appbase complements RevenueCat’s purchase infrastructure; it does not replace store billing.
A worked example: conversion is not paid conversion
Imagine 100 eligible viewers. Twelve start a trial and three buy directly. The observed paywall conversion is (12 + 3) ÷ 100 = 15%. That does not mean 15 paid customers.
If four of those trial starters later pay, trial-to-paid conversion so far is 4 ÷ 12 = 33.3%. There are seven first-time paying customers across the two purchase paths. Pending trials remain in the trial denominator, so a recent cohort’s result can change. These figures are an illustrative example, not a benchmark or an Appbase customer result.
Do not count a renewal as another new customer. Use acquisition cohorts to study new business, and transaction-date reports to study revenue collected in a period. Appbase exposes these as separate reports.
Check coverage before judging a paywall
- Are the view, transaction and provider customer linked to the same person?
- Are sandbox purchases mapped to development rather than production?
- Did billing arrive after the app event, and has the report updated?
- Are the cohorts old enough for trials to finish?
- Are missing history or unknown amounts explicitly marked unavailable?
A zero means no observed conversions in a connected, supported scope. Unavailable means there is not enough evidence to calculate the value. Those states require different actions. Appbase’s observed billing history can be partial; installing analytics today does not reconstruct every past paywall exposure.
Next, inspect the journey before the paywall and whether people return afterward. A higher trial rate is only one part of a healthy subscription app.