How Mojo grows

Originally the first paywall screen showed yearly and monthly plans side by side. The test moved the monthly plan under a 'view all plans' button so it wasn't visible at first glance, leaving the yearly plan as the default.

Added a small line next to the yearly plan stating its monthly-equivalent price (about $10/month) so it contrasts with the monthly plan (around $25). A/B tested by market.

Tested in Japan a long scrolling paywall with lots of information, social proof/reviews and a clear free-tier vs pro-tier comparison. The same design was tested in the US, where a cleaner design with a slider, background videos and punchy messages performed better.

Ran a paywall campaign for existing free users triggered on app open, with frequency capped at one paywall per week however often they open the app. Alternative trigger suggested: after a key engagement event (e.g. sharing). Checked support and did a thorough App Store review analysis for complaints. Works alongside a generous free tier.

For price tests, tracked the 7-day cancellation rate as a proxy for renewal rate. When a higher price won on new revenue, modeled a year of revenue at that price using the proxy; the higher price's cancellation rate was far above baseline, so kept the original price.

Mojo keeps a generous free tier, converts many users on day zero, and shows a paywall on app launch to retained freemium users, capped at once a week, to keep re-pitching premium features over time.

Instead of the App Store's default exchange-rate conversion of US prices, ran price tests in the top LatAm markets (Brazil, Mexico) with lower local prices.

Used ARPU in the first 7 days (ARPU7D) as the single success metric for all monetization experiments, to optimize new revenue, shorten payback and support scaling user acquisition. Seven days covered the 3-day trial plus a few days, and was a default window available in RevenueCat charts. Also monitored ARPU7D of new cohorts in RevenueCat charts after rollouts.

Used a third-party paywall platform so the PM could build and launch tests autonomously, cutting the cycle from weeks or months to days. Split new users into three geo buckets (US/Australia/Canada English, Europe, Latin America) and ran a separate test stream in each in parallel, launching on a weekly or bi-weekly cadence; each test ran 1-2 weeks until statistically significant.

When a test (usually in the US or Europe) produced a statistically significant winner, retested it in each other key geo segment and rolled it out only where it brought incremental revenue. Then watched ARPU7D of new cohorts in RevenueCat charts to confirm the lift held.

Mojo's free tier intentionally includes many features and a lot of content, less restricted than competitors, so users retain and are later converted by in-app paywall campaigns.