App Subscription Conversion
Review download-to-trial-to-paid conversion by category, region, and price, plus hard paywalls, AI-app retention, and Korean trial-rule effects using transaction logs.
Written and reviewed by Eric Han · Co-founder & CTO, reopt
Key takeaways
- Paid conversion within 35 days of download is 2.0% overall, 2.9% for health, and 1.0% for games.
- Hard paywalls show about five times freemium's download-to-paid conversion, 10.7% versus 2.1%. These are self-selected app models in paywall-vendor data. Report/blog labels differ; the interpretation follows the magnitudes.
- AI apps start more trials, 8.5% versus 5.6%, but retain fewer monthly subscribers at 12 months, 6.1% versus 9.5%.
- Of three-day trial cancellations, 55.4% occur on day one and 84% within two days.
- Some Korean Android cohorts saw trial conversion collapse after February 2025 regulation. Do not transfer pre-change or overseas values directly.
Upstream store conversion, ad costs, and retention are covered in Mobile App Acquisition and Retention.
Sample Characteristics
RevenueCat 2026 covers more than 115,000 apps and $16 billion revenue; Adapty 2026 covers 16,000 apps and $3 billion. Both use transaction logs, but only for apps using their tools; both sell paywall/subscription products (grade B). App figures do not transfer to B2B SaaS.
Categories
- Health2.9%
- Business2.6%
- Overall2.0%
- Shopping1.3%
- Games1.0%
| Category | Download → trial | Trial → paid | Download → paid within 35 days |
|---|---|---|---|
| Business | 9.1%, upper benchmark 16.2% | — | 2.6%, upper benchmark 5.0% |
| Health | 6.9%, highest performers above 23% | 37.7%, top quartile 51.4% | 2.9%, top quartile 6.2% |
| Education | 6.5% | — | — |
| Utilities | 6.5% | — | — |
| Games | 4.4% | 25.0%, top quartile 39.8% | 1.0% |
| Travel | 4.1% | 43.5%, top quartile 62.4% | — |
| Media | 4.0% | — | — |
| Photo / video | — | 22.2%, top quartile 33.1% | — |
| Shopping | — | — | 1.3% |
| Overall | — | — | 2.0% |
“—” marks chart-only values not confirmed numerically. For business, education, utilities, and media, the missing values prevent calculating end-to-end conversion by multiplying stage rates. Health gives 6.9% × 37.7% ≈ 2.6%, close to the 2.9% download-to-paid figure. Direct purchases without trials mean equality is not expected.
Regions and Price Bands
| Region | Download → trial | Trial → paid | Download → paid within 35 days |
|---|---|---|---|
| North America | 7.1% | 34.2% | 2.8% |
| Asia-Pacific | 5.7% | 31.9% | 2.4% |
| Western Europe | 5.0% | 29.7% | — |
| India / Southeast Asia | — | 15.2% | 0.7% |
| Price band | Download → trial | Download → paid within 35 days |
|---|---|---|
| High | 8.9% | 2.8% |
| Mid | 5.4% | 2.0% |
| Low | 4.4% | 1.4% |
The report's summary instead gives North America 2.6%, Western Europe 2.0%, and India/Southeast Asia 1.4%, conflicting with the table. We therefore do not use regional multipliers. Public sections contain no Korea-only figure.
Adapty's overall averages are 10.9% install-to-trial and 25.6% trial-to-paid in State of In-App Subscriptions 2026. The 26.2% in Mobile App Acquisition and Retention comes from a different report, State of Paywalls. Because RevenueCat differs, multiply stages only within one provider's table.
Paywalls and Plans
| Comparison | Figures | Notes |
|---|---|---|
| Hard paywall versus freemium, 35-day download → paid | 10.7% versus 2.1% | Report/blog labels differ; interpretation follows report magnitudes |
| 60-day revenue per install | $3.09 versus $0.38 | RevenueCat blog |
| Weekly versus annual plans | Weekly conversion 1.7–7.4 times higher | Adapty |
| Trial-start timing | 89.4% on day one | Adapty |
| Three-day trial cancellation timing | 55.4% on day one; 84% within two days | RevenueCat blog |
Much is decided on day one. Use daily time steps for app simulations.
AI Apps
Figures from the RevenueCat report:
| Metric | AI apps | Non-AI apps |
|---|---|---|
| Trial starts | 8.5% | 5.6% |
| Download → paid | 2.4% | 2.0% |
| Monthly-plan retention at 12 months | 6.1% | 9.5% |
| Annual-plan retention at 12 months | 21.1% | 30.7% |
| One-year realized LTV | $30.16 | $21.37 |
AI apps have higher conversion but lower retention than non-AI apps. Planning AI-app CAC from first-month conversion alone can misrepresent payback.
A Filing-Based Example
Duolingo Q2 2026 reports 12.7 million paid subscribers, 140.6 million MAU, and 58.7 million DAU. Paid penetration is about 9.0% of MAU or 21.6% of DAU (calculated), not a period conversion rate.
Korea
Trial Regulation
From February 14, 2025, Korea required consent for automatic transition from a free trial to paid service. RevenueCat's account says Google must obtain first-payment consent within the final 14 days of a trial, canceling automatically without it; Apple obtains consent within 30 days of the transition date.
The following RevenueCat cohorts are grouped by start date for 5–9-day trials (B).
| Start date | Android | iOS |
|---|---|---|
| 2025-01-31 | 34.5% | 41.4% |
| 2025-02-05 | 39.6% | 46.1% |
| 2025-02-07 | 1.6% | 44.6% |
| 2025-02-15 | 3.4% | 42.1% |
| 2025-02-17 | 0% | 0.4% |
The article was written on February 23, before all 5–9-day trials in the final cohort could finish (author judgment). It states that most Google Play trial expirations after February 14 were attributed to billing errors. Measure Korean trials separately before and after February 2025. Recheck implementation status before interpreting policy operationally.
Public Cases
Millie's Library's 2023 prospectus defines and reports free-subscription-to-paid conversion of 34.7–37.3% and resubscription of 76.2–87.6% for 2019–first half 2023 (internal company data, B). The denominator is people starting a free subscription. An article citing a brokerage report gives about 39% and 88% without definitions (D); use the filing instead. See more Korean Cases.
Simulation Inputs
| Stage | Input | Evidence and conditions |
|---|---|---|
| Download → trial | Health 6.9%, games 4.4%, travel 4.1% | RevenueCat, B; assumes your app is near its category median |
| Trial → paid | Health 37.7%, games 25.0%, travel 43.5% | Same source, B |
| 35-day download → paid when your position is unknown | Health mean 4.4%, | Fitted to median 2.9% and upper-quartile threshold 6.2% |
| Korea | Keep the mean and widen the interval | Separate pre/post-February-2025 regulation. Filing reference: Millie's Library 34.7–37.3%, B |
Do not use the article's approximately 39%/88% Millie's figures or pre-regulation Korean cohorts as current inputs.
App paid-conversion intervals
Choose a category to populate RevenueCat rates. This assumes your app is near the category median and excludes between-app variation. If your position is unknown, use the wider interval in Worked Scenario 4. Concentration k is an author judgment.
Each input alone moves to its 5th and 95th percentile with others at the mean. Measure the top one first.
- Download → trial423customers ~ 1,220customers
- Trial → paid553customers ~ 1,020customers
8,000 runs. Stages are independent. If they move together the real interval is wider; if they move in opposite directions it is narrower. The 90% interval holds only rate uncertainty around the expected value; the actual count adds chance variation in counts. Vertical lines mark the 5th percentile, median (darker), and 95th. The last bin collects values above the 99th percentile.
Mobile App Acquisition and Retention
Review store-view-to-install conversion, Apple Ads costs, retention, permission consent, paywalls, and new-app revenue distributions with explicit denominators.
Desktop App Conversion
Review downloadable-software funnels, Steam wishlist sales ratios, refunds, and installation warnings with explicit dates and samples.