How Weather Up grows

A/B test (50/50) with RevenueCat Experiments scheduled to start at 12:01 AM on launch day: arm A pre-selected the $40/year plan, arm B the $4/month plan; nothing else changed. Annual default: ~800 annual, ~400 monthly. Monthly default: ~1,100 monthly, ~100 annual. Judge on realized LTV and revisit after the first annual renewals (day 366).

Made a video demonstrating the new interactive widgets, originally for the paywall; it turned out well enough to become the press pitch. Sent it as a short Twitter DM ('Hey, I'm working on some really cool interactivity for widgets and weather') instead of a long email, to contacts and to journalists he did not know. Only pitch when the update is genuinely interesting enough to write about; don't send minor point releases. Kept a per-release press page with the embargo date/time.

Single tier at $4/month or $40/year (annual only ~17% off, to keep monthly revenue for monthly API costs). Price was set from his own view of the product's value and the cost of serving multi-source widgets that refresh in the background, not from customers' perceived value or competitor prices. No discount at launch, and most articles listed the $40 price.

Using Apple's newer price points, created a second annual SKU at $39.99 alongside the $40 one in App Store Connect and ran the split in RevenueCat Experiments without touching code, shipping an update or testing in sandbox.

Hard paywall with no X. When a user left the app from the paywall, a local push notification fired immediately: 'try it for seven days'.

Put an X on the hard paywall; tapping it gives a 7-day reverse trial that does not auto-convert (no payment), clearly explained; after 7 days the app locks again. Planned for Weather Up's next update, replacing the push-notification win-back.

From 2019 held a finished Apple Watch app, then home screen widgets, then waited to add interactive widgets, each time deciding the feature was no longer novel enough to get press and Apple attention on its own. Shipped everything together as Weather Up 3.0 in early 2024.

Scheduled the launch and posted the date/time as the embargo on the release's press page, squeezing it in ahead of the Apple Vision Pro launch. Apple announced its DMA compliance the same hour; some outlets published early, 9to5Mac waited a day, others skipped it. In hindsight: avoid windows when the App Store editorial team is busy and wait a month.

Built a custom SwiftUI paywall with a video of the interactive widgets behind RevenueCat's paywall footer view, which renders prices and the purchase button and handles purchase logic. Because offerings are remote, later paywall/price experiments needed no code or app update.

Weather Up ran Google AdSense ads, which showed slot-machine-style, gun and other offensive ads despite filtering, leading to bad reviews and support requests. The ads were pulled ahead of the 3.0 revamp. Because every app open costs money and the main value (widgets and Apple Watch complications) cannot carry ads, 3.0 moves to a hard paywall with a 7-day free trial.

A big 2024 update of the weather app got it featured by Apple and covered by the press. Almost no paid ads. The next year there were no updates and no advertising.

Build one or two features interesting enough for some attention, bundle them into a small relaunch (many people will see the app for the first time), and put the subscription on sale. Then keep shipping frequently and experiment with pricing and win-back offers.