Match the profile before you quote a number
Five axes that decide which benchmark column your app belongs in.
RevenueCat State of Subscription Apps 2026 · Adapty 2026
The most common benchmarking mistake is not using a stale number. It is using a real number from the wrong column. Cross-vertical medians are freemium-weighted, so they understate a hard-paywall, AI, iOS, or high-priced app, sometimes by a factor of five. Before you compare your funnel to anything, place your app on these five axes and read the matching row.
1. Access model. The single biggest swing.
| Access model | p10 | p50 | p75 | p90 |
|---|---|---|---|---|
| Hard paywall (no free path) | 4.2% | 10.7% | 20.0% | 38.7% |
| Freemium / soft (a free path exists) | 0.3% | 2.1% | 4.5% | 8.2% |
Hard converts about 5x freemium at the median. The cross-vertical "about 2%" figure you see quoted everywhere IS the freemium number. A gated free tier with an early paywall is still freemium: benchmark it at the upper freemium range, not the hard-paywall median.
2. AI or not.
| Stage | AI apps | Non-AI / average |
|---|---|---|
| Install to trial | 5.3% (Adapty) / 8.5% (RevenueCat) | 5.6% (RC) / ~10.9% (Adapty avg) |
| Trial to paid | 20.6% | 25.6% avg |
| Install to paid (D35) | 2.4% | 2.0% |
| Y1 revenue per payer | $30.16 | $21.37 |
AI apps have a top-of-funnel problem, not a monetization problem. They start trials about half as often, then convert and monetize comparably once subscribed, carrying roughly 41% higher revenue per payer. If you run an AI app, the lever is install-to-trial: presentation, trust, and comprehension at the paywall. It is not trial-to-paid.
3. Platform.
| Platform | Weekly | Monthly | Annual | D35 (RC) |
|---|---|---|---|---|
| iOS | 1.55% | 0.49% | 0.53% | 2.6% |
| Android | 0.26% | 0.17% | 0.14% | 0.9% |
iOS converts about 3x Android overall, and about 6x on weekly plans. Blended numbers hide this. The practical consequence is on the buying side: iOS can support roughly 3 to 6x higher CAC per install.
4. Paywall placement.
| Placement | With trial | No trial |
|---|---|---|
| Onboarding | 1.35% | 0.82% |
| In-app | 0.89% | 0.76% |
Onboarding placement plus a trial nearly doubles in-app-with-no-trial. The trial effect is largest at onboarding placement, which is why the two decisions should be made together rather than separately.
5. Category.
Only after the four axes above. Category narrows the comparison, it does not set it. Pull the download-to-trial and D35 rates for your vertical, then chain through the matching trial-to-paid rate rather than mixing a vertical rate with a cross-vertical one.
Where this stops
These axes cut funnel, retention, and revenue benchmarks. They do not cut paid social. No paid-social publisher reports by access model, AI status, platform, or paywall placement, and the vertical taxonomies they do publish contain no subscription-app row. If you are benchmarking paid social, match to the closest published industry, say out loud that you substituted, and keep the caveat attached to the number.
Worked example
An AI journaling app with a gated free tier, an onboarding-placed 14-day trial, iOS-first. Read it as: freemium access model, so roughly a 2% D35 floor with an upper-freemium target around 4 to 5%. Apply the AI penalty on trial starts. Apply the iOS uplift. Credit the strong onboarding-plus-trial placement, worth about 1.35% on placement alone. Blend those into a target band. Do not quote the 10.7% hard-paywall median to an app that keeps a free path, which is the single most common version of this error.
The honest limit: every number here is an industry benchmark, not a diagnosis. A benchmark tells you where you sit. It does not tell you why, and it does not rank what to fix.
Want this run on your app?
This is the selector layer. The full version places your own numbers against these columns, metric by metric, with the source, year, and caveat on every row, and names your two or three biggest gaps with a next step each.
Request a Benchmark Comparison