Skip to content
Konstantin Khalyushev
ENRUESPT

Churn and refunds

Refund rate

Of paid subscriptions, the share refunded during the first billing period.

Denominator
paid subscriptions
Cohort
a subscription cohort
Time horizon
the first billing period
Sources
3
Observations
11

Compare your metric

Each source is compared against independently. Nothing here is averaged across sources, and there is no combined industry figure. This runs entirely in your browser. Your figures are never sent anywhere, never stored, and never included in analytics — only which metric you chose is counted.

View the benchmark data

Can these be compared?

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

These observations cannot be combined into one figure, so they are shown separately. Different sub-segments: not broken out vs low-priced apps.

RevenueCat — State of Subscription Apps 2026

Published 2026-03-19 · 7 observations

Sample
Over 115,000 apps using RevenueCat for in-app subscription management, covering more than $16 billion in revenue and more than a billion transactions. Apps must have active subscription revenue and meet a minimum install or revenue threshold.
How it is aggregated
App-level metrics aggregated across apps and reported as medians and quartiles (Q1/P25, median/P50, Q3/P75, P90). Each app is a data point regardless of size.
Measurement period
Target time frame 2025; older data used where a metric needs it (e.g. third annual renewal rates).
Denominator
paid subscriptions

3%–4%

Median

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“Most categories cluster between 3-4% refund rate.”

RevenueCat — State of Subscription Apps 2026Published 2026-03-19Billing health · Refund rate by app category

2.7%

Median

  • Low-priced apps

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“Low-priced median: 2.7%.”

RevenueCat — State of Subscription Apps 2026Published 2026-03-19Billing health · Refund rate by pricepoint

3.9%

Median

  • Mid-priced apps

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“Mid-priced median: 3.9%.”

RevenueCat — State of Subscription Apps 2026Published 2026-03-19Billing health · Refund rate by pricepoint

4.5%

Median

  • High-priced apps

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“High-priced median: 4.5%.”

RevenueCat — State of Subscription Apps 2026Published 2026-03-19Billing health · Refund rate by pricepoint

3.4%

Median

  • North America

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“North America shows tightest distribution (3.4% median, 14.2% max) — most predictable market.”

RevenueCat — State of Subscription Apps 2026Published 2026-03-19Billing health · Refund rate by geography

4.2%

Median

  • AI apps

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“AI median: 4.2%; if you're below ~2.2%, you're in the bottom quartile.”

RevenueCat — State of Subscription Apps 2026Published 2026-03-19AI vs. non-AI · Refund rate by AI vs. non-AI

3.5%

Median

  • Non-AI apps

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“Non-AI median: 3.5%; best performers reach ~1% for both segments.”

RevenueCat — State of Subscription Apps 2026Published 2026-03-19AI vs. non-AI · Refund rate by AI vs. non-AI

Adapty — State of In-App Subscriptions 2026

Published 2026-01-01 · 1 observation

Sample
Subscription data from more than 16,000 apps representing $3 billion in subscription revenue.
How it is aggregated
Conversion benchmarks are computed as app-level rates (trials or purchases divided by installs) within a segment, then aggregated as the AVERAGE of per-app rates. Medians and percentiles are reported alongside averages on some charts.
Measurement period
2025, with 2023 comparisons on selected charts.
Denominator
paid subscriptions

8.3%

Average

Evidence One of sample, statistic or definition is missing or ambiguous.

Methodology
Computed as app-level rates within the segment, then aggregated as the average of per-app rates. Averages of per-app rates are pulled upward by strong outliers, so this sits above a median of the same population. The report does not state whether the denominator is limited to the first billing period, as RevenueCat's refund rate is.
Comparability
Roughly twice RevenueCat's 3-4% median. Some of that gap is mean-vs-median and some may be a different denominator, so the two should not be differenced.
As published
“8.3% Refund”

Adapty — State of In-App Subscriptions 2026Published 2026-01-01Conversion highlights · Conversion rates by user journey stage (All Categories, Global)

RevenueCat — State of Subscription Apps 2025

Published 2025-04-01 · 3 observations

Sample
Apps using RevenueCat for in-app subscription management, segmented by category, platform, region, price point and engagement strategy.
How it is aggregated
App-level metrics reported as bottom quartile, median, upper quartile and top 10% of app performance.
Measurement period
Primarily 2024.
Denominator
paid subscriptions

4.86%

Median

  • Education

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“Education and health & fitness apps have the highest refund rates, at 4.86% and 4.71% respectively.”

RevenueCat — State of Subscription Apps 2025Published 2025-04-01Billing health · Refund rate by category

4.71%

Median

  • Health & Fitness

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“Education and health & fitness apps have the highest refund rates, at 4.86% and 4.71% respectively.”

RevenueCat — State of Subscription Apps 2025Published 2025-04-01Billing health · Refund rate by category

1.51%

Median

  • Travel

Evidence Sample, statistic and metric definition all stated by the publisher.

As published
“Travel apps have the lowest refund rate (1.51%)”

RevenueCat — State of Subscription Apps 2025Published 2025-04-01Billing health · Refund rate by category

Pairwise comparability

The verdict below is produced by a fixed set of rules, not a model. It looks at the metric definition, the denominator, the cohort, the statistic, the segment, the dataset and the publication dates, and reports every reason it found.

RevenueCat 2026 3%–4%Adapty 2026 8.3%

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

  • One is a median, the other a average. In a skewed distribution — and app performance is heavily skewed — a mean sits above the median, so the gap between these numbers partly reflects the statistic rather than the products.
  • Different datasets: RevenueCat (115,000+ apps, $16bn in revenue) versus Adapty (16,000+ apps, $3bn in subscription revenue). Each publisher sees only its own customers, so the app populations differ.
  • Different aggregation methods. RevenueCat 2026: medians and quartiles across apps. Adapty 2026: the average of per-app rates.

RevenueCat 2026 2.7%Adapty 2026 8.3%

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

  • One is a median, the other a average. In a skewed distribution — and app performance is heavily skewed — a mean sits above the median, so the gap between these numbers partly reflects the statistic rather than the products.
  • Different sub-segments: low-priced apps vs not broken out.
  • Different datasets: RevenueCat (115,000+ apps, $16bn in revenue) versus Adapty (16,000+ apps, $3bn in subscription revenue). Each publisher sees only its own customers, so the app populations differ.
  • Different aggregation methods. RevenueCat 2026: medians and quartiles across apps. Adapty 2026: the average of per-app rates.

RevenueCat 2026 3.9%Adapty 2026 8.3%

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

  • One is a median, the other a average. In a skewed distribution — and app performance is heavily skewed — a mean sits above the median, so the gap between these numbers partly reflects the statistic rather than the products.
  • Different sub-segments: mid-priced apps vs not broken out.
  • Different datasets: RevenueCat (115,000+ apps, $16bn in revenue) versus Adapty (16,000+ apps, $3bn in subscription revenue). Each publisher sees only its own customers, so the app populations differ.
  • Different aggregation methods. RevenueCat 2026: medians and quartiles across apps. Adapty 2026: the average of per-app rates.

RevenueCat 2026 4.5%Adapty 2026 8.3%

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

  • One is a median, the other a average. In a skewed distribution — and app performance is heavily skewed — a mean sits above the median, so the gap between these numbers partly reflects the statistic rather than the products.
  • Different sub-segments: high-priced apps vs not broken out.
  • Different datasets: RevenueCat (115,000+ apps, $16bn in revenue) versus Adapty (16,000+ apps, $3bn in subscription revenue). Each publisher sees only its own customers, so the app populations differ.
  • Different aggregation methods. RevenueCat 2026: medians and quartiles across apps. Adapty 2026: the average of per-app rates.

RevenueCat 2026 3.4%Adapty 2026 8.3%

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

  • One is a median, the other a average. In a skewed distribution — and app performance is heavily skewed — a mean sits above the median, so the gap between these numbers partly reflects the statistic rather than the products.
  • Different geographies: North America vs not broken out. Publishers define their regions differently, so similar-sounding names are not the same country list.
  • Different datasets: RevenueCat (115,000+ apps, $16bn in revenue) versus Adapty (16,000+ apps, $3bn in subscription revenue). Each publisher sees only its own customers, so the app populations differ.
  • Different aggregation methods. RevenueCat 2026: medians and quartiles across apps. Adapty 2026: the average of per-app rates.

RevenueCat 2026 4.2%Adapty 2026 8.3%

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

  • One is a median, the other a average. In a skewed distribution — and app performance is heavily skewed — a mean sits above the median, so the gap between these numbers partly reflects the statistic rather than the products.
  • Different sub-segments: AI apps vs not broken out.
  • Different datasets: RevenueCat (115,000+ apps, $16bn in revenue) versus Adapty (16,000+ apps, $3bn in subscription revenue). Each publisher sees only its own customers, so the app populations differ.
  • Different aggregation methods. RevenueCat 2026: medians and quartiles across apps. Adapty 2026: the average of per-app rates.

RevenueCat 2026 3.5%Adapty 2026 8.3%

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

  • One is a median, the other a average. In a skewed distribution — and app performance is heavily skewed — a mean sits above the median, so the gap between these numbers partly reflects the statistic rather than the products.
  • Different sub-segments: non-AI apps vs not broken out.
  • Different datasets: RevenueCat (115,000+ apps, $16bn in revenue) versus Adapty (16,000+ apps, $3bn in subscription revenue). Each publisher sees only its own customers, so the app populations differ.
  • Different aggregation methods. RevenueCat 2026: medians and quartiles across apps. Adapty 2026: the average of per-app rates.

RevenueCat 2025 4.86%Adapty 2026 8.3%

Roughly comparable

The same thing is being measured, but at least one condition differs. Read the direction, not the gap.

  • One is a median, the other a average. In a skewed distribution — and app performance is heavily skewed — a mean sits above the median, so the gap between these numbers partly reflects the statistic rather than the products.
  • Different category scope: Education vs not broken out. An all-categories figure is not a substitute for the category you are in.
  • Different datasets: RevenueCat (apps on RevenueCat, sample size not stated) versus Adapty (16,000+ apps, $3bn in subscription revenue). Each publisher sees only its own customers, so the app populations differ.
  • Different aggregation methods. RevenueCat 2025: quartiles and medians across apps. Adapty 2026: the average of per-app rates.
Editions over time

Each edition of a report is stored separately, and older observations are never overwritten. A series is only shown as a trend when the underlying method held still.

RevenueCat — State of Subscription Apps

  1. Edition 2025

    2025-04-01

    • 4.86%MedianEducation
    • 4.71%MedianHealth & Fitness
    • 1.51%MedianTravel

  2. Edition 2026

    2026-03-19

    • 3%–4%Median
    • 2.7%MedianLow-priced apps
    • 3.9%MedianMid-priced apps
    • 4.5%MedianHigh-priced apps
    • 3.4%MedianNorth America
    • 4.2%MedianAI apps
    • 3.5%MedianNon-AI apps

Churn and refundsCompare your productMethodology and sourcesAll benchmarks