2026年订阅应用市场状况报告

built on the world’s largest in-app subscription data set. as consumers find more solutions to their problems inthe app stores. On the demand side, if AI ushers in anunprecedented era of productivity and consumer surplus,it’s inevitable that some of that surplus will be spent in theapp store. At least we hope.We built SOSA 2026 from over 115,000 apps, $16 billion Three years ago, about 2,000 newsubscription apps launched every month.Today that number is almost 15,000.AI removed a decade old supplyconstraint on apps, and now we’re goingto have to process this glut of appsas consumer demand most likely lags.This will be seen as more competition,higher CACs, and higher churn.But the silver lining is that I think this is just a shock. New apps in revenue, and more than a billion transactions. We slicedit by category, platform, trial length, paywall strategy, AIvs. non-AI, and probably a few dimensions we’ll forget weincluded until someone tweets about them. It’s a lot.We know.The subscription app market is bigger, faster, and more unforgiving than it’s ever been. But the patterns are thereif you look. The operators who study the data, and learnthe tools, are the ones who will come out ahead. My hopeis this report helps you be one of them. are being invented with new capabilities never seen before.The relatively low cost of trying software is allowing appsto be built for niches never before economically viable. Thissmorgasbord of software will, in itself, incur more demand with a larger segment to avoid the possibility of inferring anysingle app’s data. Throughout this report, numbers representaggregated totals, averages, medians, quartiles, or othersummary statistics. No app-specific or developer-specificdetails are ever disclosed. Overview of the datasetThis report draws on subscription app performance data from a wide range of apps that use RevenueCat’s platform.Our goal is to provide a comprehensive snapshot of howapps are performing under different scenarios, across iOS,Android and web ecosystems.→Scope of apps included:we included apps that have Statistical definitionsThroughout this report we aim to use clear, accessible active subscription revenue, meet a minimum thresholdof installs or revenue (to ensure statistically meaningfulfindings), and have integrated RevenueCat for in-appsubscription management.→Time frame:the target time frame for metrics in this report language and minimize unnecessary jargon. However, somekey statistical terminology is used at times.The following measures of central tendency and spread have been used to illustrate app performance.→Bottom quartile (Q1):The value below which 25% of the is 2025. In some cases we have pulled older data to runcertain calculations (we can’t calculate third renewal ratefor annual subscriptions bought in 2024, for example)→Size and composition:we analyzed over 115,000 apps dataset falls. An app that falls into the bottom quartile isamong the lower 25% of performers on that metric.→Median (Q2):The middle value, with half of the data above and half below. When comparing your own metrics to themedian, you can see if you are performing above or belowthe midpoint of the industry.→Upper quartile (Q3):The value above which 25% of thedataset falls. An app in the upper quartile is among the top apps across all app categories, covering more than$16 billion in revenue across more than a billiontransactions. The apps vary in scale, from indie teamsto mid-size organizations and large publishers.→Revenue channels:the dataset includes both apps thatprimarily generate revenue from in-app subscriptions 25% of performers for that metric.→P90:The 90th percentile. This indicates the point at orabove which 10% of the dataset lies. An app at P90 is and those that generate a portion of revenue fromsubscriptions alongside other revenue channels.→Anonymity:all data is anonymized and aggregated,ensuring that no single app’s performance metrics are outperforming 90% of apps in that particular metric (aka,crushing it).Reading the charts individually identifiable. The findings are presented asaggregated performance benchmarks across segments,categories, and platforms.Anonymization and data privacyTo preserve the confidentiality of individual apps, we apply to show the distribution of a given performance metric(e.g. RPI, LTV) across apps. The ‘box’ represents the bulk of distribution, the interquartile range, while the ‘whiskers’represent the lower and upper bounds of performance. Methodologycontrols to ensure that if a segment has too few apps,results are either omitted from the report or combined by total installs. This metric highlights how well an appmonetizes each new user. Percentile mapping:→Lower ‘whisker’:marks P10 (10th percentile), the bottom 10% of app performance→Bottom of the ‘box’:marks P25 (25th percentile) andrepresents Q1 (bottom quartile) — apps below this makeup the lowest 25% of performers→Marker inside the ‘box’:marks P50 (50th perc