November 2026 Summaries

No summary generated yet.

Apple’s Subscription Bundles and Suites, introduced at WWDC26 and currently available through early access, let developers package auto-renewable subscriptions across apps, but they serve different business models. Bundles combine up to five separately purchasable subscriptions with matching durations at a discounted price and can include apps from one or as many as five developers, supporting cross-promotion, prorated upgrades, and the option for customers to retain individual subscriptions after leaving a bundle. Suites provide one subscription across up to 15 related apps from a single developer, but their included entitlements cannot be sold separately, making them best suited to unified app collections. Developers must implement StoreKit 2, ensure support on iOS, iPadOS, macOS, or tvOS 27 or later, submit an App Store Connect request with participating app identifiers, and complete Apple onboarding; multi-developer participants must each apply and sign an addendum. Apple will coordinate pricing and launch details after approval, although public information has not yet clarified how revenue sharing and payments work for multi-developer bundles.

Privacy changes such as Apple’s ATT and SKAdNetwork reduced deterministic mobile attribution, prompting subscription apps to adopt more first-party data, modeled attribution, cohort analysis, incrementality testing, and marketing mix modeling, but these expanded systems often create disagreement rather than alignment. Different platforms and teams can produce legitimately different metrics because they answer different questions, so the goal should not be to force dashboards to match but to establish each system’s normal reporting gap and investigate meaningful changes in that gap. Effective measurement begins with recurring business decisions, matching each one to a fast signal, downstream guardrail, business-level anchor, and evidence threshold proportional to the cost and reversibility of being wrong. Directional platform data can support rapid, low-stakes optimizations, while pricing changes, major budget reallocations, and financial planning require stronger evidence such as controlled experiments, cohort performance, incrementality, and realized financial data. Blended metrics, including total spend, acquired customers, and revenue, provide an essential economic reality check against which attribution claims can be evaluated, although averages must be supplemented by segmentation to reveal differences among audiences, markets, plans, and cohort maturity. Good measurement therefore relies on understanding what each metric can and cannot answer, using blended outcomes to assess overall business health, and applying increasingly rigorous evidence to higher-stakes decisions.

Mike Charniatsou of women’s health app Flo argues that subscription growth should reflect users’ changing needs over time rather than focus solely on immediate paywall conversion, with 40% of Flo subscriptions occurring after six months as users encounter offerings related to fertility, pregnancy, or perimenopause. He emphasizes that conversion is a lifecycle involving trial activation, retention, and long-term value, so experiments that improve short-term metrics may still harm the business through effects such as increased refunds. Flo evaluates experiments over extended periods and distinguishes evidence from assumptions: a perimenopause onboarding step initially reduced conversion twice, but analysis revealed that users could not answer its questions; revisions based on this finding ultimately produced a 23% uplift that was confirmed through a further test. The company also uses marketing performance to identify emerging user needs, such as sex-life content that led to a successful tailored onboarding branch, while sharing insights across teams. Charniatsou cautions against copying successful tactics from other apps, noting that approaches such as Blinkist-style trial reminders failed at both Flo and his previous company, and advocates understanding each product’s underlying user problem before adapting an idea.

An investigation into delayed subscription-plan selection in RevenueCat’s Android Compose paywalls found that the main cause was excessive invisible recalculation on the UI thread rather than background blur or visible recomposition. Because all components eagerly read selection-related state while resolving conditional style overrides, a package tap invalidated derived state across the entire paywall, forcing hundreds of override evaluations and variable-template passes even for components unaffected by the selection. The team reproduced the issue with instrumented tests, added performance counters and feature switches, and reduced latency by making condition inputs lazy so only relevant state is read, caching regular expressions used in accessibility text processing, and adding value-based equality for downloaded fonts to avoid unnecessary text recomposition and repeated main-thread font loading. Additional improvements cached Markdown, shadows, and window-size calculations, moved button alpha animation work into the draw phase, stopped an invisible indeterminate spinner from continuously rendering frames, and clarified that simply reordering synchronous listener calls cannot allow a new frame until the current main-thread task finishes. The work emphasizes that derivedStateOf can prevent recomposition without preventing costly recalculation, making it important to limit state reads to paths that actually need them and validate suspected performance causes with controlled before-and-after measurements.

RevenueCat redesigned its AI paywall editor after finding that its original JSON-based agent could complete requested edits but often produced visually weak designs despite growing adoption and prompt, model, and screenshot-based improvements. The team concluded that forcing models to translate design intent into proprietary JSON manipulation tools lost capabilities the models had developed through training on familiar UI coding formats such as JSX. They therefore introduced a workspace-based coding-agent approach in which paywalls are converted from JSON to editable JSX, validated by a strict compiler, and transformed back into the JSON required by the backend and SDKs. The revised system uses familiar file operations, lazily loaded component types, reusable snippets, localization files, and app context, while requiring a shorter system prompt and fewer custom skills. Following offline tests, production-trace comparisons, human blind evaluations of design quality, infrastructure upgrades, and stronger reliability mechanisms, the editor was fully deployed on September 2. Across roughly 46,000 requests, positive feedback increased 83% and negative feedback declined 40%, with the largest improvements occurring in creating or substantially redesigning paywalls, although the new approach introduced greater latency.

Apple Retention Messaging, available to all developers through App Store Connect without code or server work, displays a short value-focused message when subscribers initiate cancellation, with optional images or eligible retention offers. App Store Connect provides a simple setup but limits each subscription to one universal message and restricts offers to subscribers with at least one month of paid history, while Apple’s Retention Messaging API, supported by RevenueCat, enables real-time targeting, personalized messages, plan switches, trial-user offers, translations, and A/B testing. Effective messages should reflect why users subscribe and cancel, combine clear feature reminders with emotional benefits, and use discounts selectively because price is not always the main churn driver and higher save rates may not produce greater revenue. Developers can configure messages, localizations, images, offers, eligibility criteria, and subscription assignments in App Store Connect before Apple review and sandbox testing, then evaluate outcomes through save rate, offer acceptance, continued renewal, churn, and realized lifetime value. RevenueCat’s real-time integration is positioned as an option for teams seeking more granular subscriber segmentation, experimentation, multilingual management, and post-cancellation revenue measurement.

Rhys Kentish, a London-based mobile engineer, created Touch Grass as a screen-time app that locks distracting apps until users photograph real grass, initially to curb his own scrolling habits. After prototype posts on Reddit, Hacker News, and X attracted major attention, the app accumulated about 40,000 preorders and reached roughly 50,000 active users at launch, but its original Google Vision-based image verification system would have cost an estimated $20,000 per week, forcing Kentish to rebuild the technology for on-device processing immediately before release. The app’s simple, humorous premise and concise marketing—often a single sentence paired with a screen recording—helped it spread across several platforms, while Kentish’s Tinder parody app, Tender, gained separate publicity after Tinder objected to the use of the trademarked word “swipe” in its App Store listing. Despite Touch Grass’s viral launch, Kentish identified retention as its main challenge because users can easily uninstall it when they want access to blocked apps, prompting a redesign with game-like rewards, collectible critters, and pet-inspired guides. The discussion also emphasizes using preorders to capture viral interest, treating copycats as evidence of traction rather than a reason to avoid marketing, and building a personal brand and distinctive character around an app to make it harder to replicate.

App Growth Annual 2026 brought more than 400 in-person and 5,000 virtual app builders to New York for presentations, workshops, product announcements, and networking focused on app growth and monetization. Speakers highlighted contrarian strategies such as Babbel’s premium-first model and lifetime plans, Burner’s emphasis on returning users rather than preventing inevitable churn, and Tolan’s hybrid subscriptions, top-ups, and tiered pricing to offset AI compute costs. Onboarding discussions stressed that the appropriate amount of friction depends on an app’s complexity and stakes, while paywall and lifecycle messaging sessions emphasized clearly communicating trial terms, demonstrating value before payment, and using contextual retention messaging. RevenueCat announced initiatives including Early Payouts, ad monetization, an AI advisor called Rico, and Targeting by Checkpoint, while noting that app launches have risen faster than consumer spending and that newer app cohorts are growing more rapidly than older ones. The event also featured 36 workshops, a subscription-app-themed game show, and an after-party performance by Marc Rebillet.

App Growth Annual 2026 has released six previously exclusive virtual workshops for free, including slides, handouts, and additional reading, following an event that drew more than 5,000 online registrants. The sessions cover customer-language-based copywriting, user research methods behind organic app growth, AI-assisted competitor-ad analysis and creative production, a four-stage framework for validating demand and scaling app economics, AI workflows for managing Meta advertising, and a closed-loop system for generating and evaluating AI ad creative. Across the workshops, speakers emphasize that AI can speed up research, reporting, copywriting, and production, but effective results depend on reliable customer insights, clear brand and product context, structured testing, and human judgment. Viewers can also access the event’s main-stage keynote recap featuring speakers from Babbel, Burner, Tolan, Klarna, and related growth experiments.