Enforcement Reality: Apple Still Draws Hard Lines
Apple's brief removal of Cal AI — the MyFitnessPal-owned calorie tracker — underscored that the company is actively policing how developers implement external payment systems, even post-Epic ruling. Cal AI violated multiple guidelines: it bypassed wiki:in-app-purchase by embedding Stripe directly into checkout, removed Apple's IAP option entirely, and used deceptive billing patterns that obscured renewal terms. The app also employed what Apple termed "manipulative tactics," including a secondary subscription prompt after users dismissed the first offer — a tactic that crossed the line into dark patterns rather than legitimate wiki:conversion-rate-optimization-cro.
The underlying rule remains unchanged: U.S. developers may link to external payment systems, but they must still offer Apple's IAP alongside that link. The only exceptions are "reader" apps (books, audio, video streaming). Cal AI did not qualify. After addressing the violations, the app returned to the store, but the episode serves as a reminder that the App Store wiki:app-review team is watching how developers structure checkout flows, especially when those flows generate significant revenue. Apple is willing to lose its commission share from a viral app if it means maintaining control over the rules.
A resurfaced 2020 email exchange — in which Apple executive Eddy Cue suggested raising subscription commissions from 30% to 40% — offers historical context. At the time, Apple believed it was leaving money on the table. Today, regulatory pressure and antitrust scrutiny have pushed in the opposite direction, but the Cal AI enforcement shows that Apple's priority is not maximizing short-term revenue share but rather preserving the structural integrity of its payment infrastructure.
Freemium Tier Design: The Difference Between 2% and 50% Conversion
Freemium models carry lower median conversion rates — around 2.1% compared to 10.7% for hard paywalls — but the performance spread is enormous. The strongest freemium apps, including Duolingo, Slack, and Strava, achieve 42–58% retention rates after one year, far outperforming both median freemium and many hard paywall products. The difference is execution, specifically how developers define what belongs in the free tier versus the paid tier.
The first step is clarifying the strategic purpose of freemium. Is the goal to stand out in a competitive market? Drive organic growth through word-of-mouth? Build a habit that requires time to form? Generate volume because the product needs data or network effects to function? Each goal shapes how generous the free tier should be. Apps entering crowded markets (like Healthi against Weight Watchers) often start with a very generous free tier to gain traction, then tighten it over time. Apps driven by word-of-mouth (like Duolingo, where 80% of users arrive organically) treat free users as the primary distribution channel, not as freeloaders.
The second step is defining what user success looks like and working backwards to the actions that predict it. Free users should be able to achieve meaningful progress toward their goal — but not reach the full solution. This requires mapping each feature to free, paid, or limited, with a written strategic rationale. Strava, for example, keeps GPS activity recording fully free because every tracked activity generates content for the social feed and strengthens the network. Segment leaderboards are limited to personal-only for free users; full leaderboards are paid. Competitive athletes need to see where they rank, which creates a natural upgrade trigger.
The third step is choosing a freemium architecture: taster (same product with usage limits), split (different features for different user types), or hybrid (both). Loom uses the taster model — five-minute recordings, 25 videos max. Casual users stay free; serious users hit the limit and upgrade. CapCut uses the split model — basic editing is free, advanced tools are paid. Slack and ChatGPT use hybrid models, combining usage limits with exclusive premium features.
Critically, freemium apps should not default to lower pricing. Higher-priced freemium apps ($10+/month) actually convert better than low-priced ones (2.8% vs. 1.4% at Day 35). If a user has experienced genuine value in a free tier and still chooses to upgrade, they are a high-intent buyer. A generous free tier should give you confidence to charge more, not less.
Finally, high-performing freemium apps define a "Bill of Rights" for free users — an internal document that acts as a constitution for what cannot be moved behind the paywall. Life360, used by over 80 million active users, established that the core map, location history, and place alerts must always be free. No A/B test is permitted to gate those features, even if conversion rates temporarily lift. The cumulative effect of chipping away at the free tier over time is almost impossible to measure until word-of-mouth collapses. Nobody runs a holdout group long enough to see that decay.
Exit Offers and Dynamic Paywalls: Catching Users at the Point of Abandonment
Exit offers function like e-commerce cart abandonment recovery: when a user dismisses a paywall without purchasing, a second offer appears automatically, typically a lower price, longer trial, or different plan. The tactic works, but only if the paywall is presented using a method that controls the dismiss flow.
On Android, exit offers require PaywallDialog or PaywallActivityLauncher (for Activity-based flows). The embedded Paywall composable does not support exit offers because it has no concept of "dismiss" — it is just a composable in a navigation graph. When a user taps close, the navigation action happens immediately; there is nothing to intercept. Switching from the embedded Paywall to PaywallDialog is straightforward and does not alter the user experience, but it is a prerequisite for exit offers.
Exit offers are configured entirely in the RevenueCat dashboard. Developers create a separate offering for the exit offer (e.g., a discounted monthly plan), attach a paywall to it, and link it to the main paywall. The SDK preloads the exit offering data in the background, intercepts the dismiss action, and presents the exit offer before closing. If the user has already completed a purchase on the main paywall, the exit offer does not trigger.
On iOS, exit offers work with presentPaywall() and presentPaywallIfNeeded() but not with PaywallView embedded in SwiftUI. The same structural limitation applies. Apple's App Store Review Guideline 5.6 (Developer Code of Conduct) has been cited in some rejections for apps that show additional offers when users try to dismiss a paywall, framing it as a "manipulative practice." Enforcement is inconsistent; some apps use exit offers without issues, while others are rejected. A safer starting approach is to enable exit offers only for Android users via RevenueCat's Targeting feature.
RevenueCat's new Paywall Rules feature extends this level of control to individual paywall components. Developers can now show or hide components based on preset conditions (e.g., whether a trial is available, which package is selected) or custom variables, all without releasing a new app version. A single paywall can now handle multiple scenarios: trial timelines appear only when trials are offered, package options swap based on user attributes, messaging adjusts for promotional offers. Rules are evaluated at runtime in a defined priority order, and components can be governed by multiple rules simultaneously.
This reduces the need to maintain separate paywall variants for different user segments. Instead of building five paywalls to cover five edge cases, you build one paywall with conditional logic. The result is faster iteration, cleaner codebases, and more precise targeting.
India's App Market: Record Revenue, But Not for Local Players
India's mobile app market crossed $300 million in in-app purchase revenue during Q1 2024, up 33% year-over-year. Non-gaming apps drove the growth, generating over $200 million (up 44% YoY) and increasing their share of overall spending. The gains were concentrated in utilities, video streaming, and generative AI.
The broader trajectory is clear: annual IAP revenue has risen from $520 million in 2021 to over $1 billion in 2025, with projections reaching $1.25 billion this year. Downloads have stabilized around 25 billion annually, but time spent on apps continues to grow, signaling deeper engagement and greater willingness to pay for digital services.
Yet much of that spending is captured by global platforms. Google One, Facebook, ChatGPT, and YouTube ranked among the top earners. Domestic players were more prominent in video streaming (JioHotstar, SonyLIV), but the download leaderboard followed a similar pattern: ChatGPT, Instagram, and the Chinese short-drama app FreeReels led, followed by Indian apps including Story TV, JioHotstar, and Meesho.
India remains a relatively low-spending market despite record revenue. The market generates about $0.03 in revenue per download, compared to more than $0.20 in Southeast Asia and Latin America. The issue is not user volume or engagement; it is conversion mechanics. Spending remains concentrated in mature segments — productivity, social media, video streaming — which account for about half of the top 10 revenue-generating apps. Video streaming alone commands outsized monetization power.
Emerging categories offer room for growth. Generative AI apps saw downloads rise 69% year-over-year, with ChatGPT ranking among the top apps by both installs and revenue. India has long been ChatGPT's largest market by users. Short-drama platforms grew even faster, with downloads up more than 400%, led by FreeReels. The trend suggests India's app market is early in its monetization journey, even as new categories expand what users are willing to pay for.
The challenge for Indian developers is not demand; it is execution. Global platforms dominate because they have mastered wiki:conversion-rate-optimization-cro, paywall optimization, and retention rate at scale. Local players who want to capture more of the $1+ billion in annual IAP revenue need to close that execution gap. The market is growing, but the gains will continue flowing to whoever builds the tightest funnel.
The Mechanics Are What Separate Winners from Also-Rans
Simply Ltd's portfolio — Simply Piano, Simply Guitar, Simply Draw, Simply Sing — generates over $3 million per month from in-app purchases alone, with additional millions flowing through web-to-app subscription funnels. The formula is straightforward but non-trivial to execute: create a genuinely valuable product focused on one core use case, deliver fast value during the first lesson, personalize that lesson based on user experience level, and lock subsequent lessons behind a paywall only after users have experienced real progress.
The first lesson is completely free and interactive. High-quality video production accompanies every step. By the time users reach the paywall, they have already made progress, trust the app's ability to deliver, and are primed to start a trial or pay. The paywall itself is not a static screen; it includes a long audio voiceover that walks through the value proposition, reinforcing the decision.
This approach mirrors the broader theme: monetization precision is not about tricks or hacks. It is about understanding where users experience value, structuring the funnel to reinforce that value, and presenting the ask at the moment when the decision feels natural. Whether that is a freemium tier designed around user intent, an exit offer that catches abandonment, or a dynamic paywall that adapts to context, the mechanics are what separate high performers from median apps.
Platform enforcement remains strict, tooling continues to advance, and markets like India are maturing rapidly. The opportunity is there. Execution is the filter.