AI-powered apps can make money, but struggle with long-term retention, new data shows

With the apical app stores flooded with AI apps, developers whitethorn deliberation the champion stake for turning a nett is to integrate artificial quality exertion into their ain products. However, a caller survey focused connected the subscription app ecosystem crossed iOS, Android, and web is calling that presumption into question.

RevenueCat, a institution that offers subscription absorption tools utilized by implicit 75,000 app developers, said successful its 2026 State of Subscription Apps Report that AI integration is not a warrant of semipermanent retention. Instead, AI-powered apps conflict to clasp subscribers, with radical canceling their yearly subscriptions — a metric known arsenic churn — 30% faster than non-AI apps, astatine the median, according to the report.

The study is based connected an investigation of the subscription app providers that usage RevenueCat’s tools to negociate their much than 1 cardinal in-app transactions, generating much than $11 cardinal successful gross for developers annually. As 1 of the much fashionable tools successful this space, its information represents a steadfast illustration successful presumption of inclination analysis.

Among the galore absorbing findings, the study noted that astir of the apps utilizing the company’s level are not yet powered by AI. AI-powered apps relationship for 27.1% of apps crossed each categories, compared with 72.9% for non-AI apps. Still, it’s a increasing category, arsenic astir 1 successful 4 apps is present AI-powered.

(To beryllium clear, the AI-powered apps class doesn’t lone see the fashionable AI chatbots, similar ChatGPT and Gemini, but besides includes immoderate app that markets itself arsenic being AI-powered.)

REvenuecat: AI vs Non-AI apps by categoryImage Credits:RevenueCat

Photo & Video apps person the biggest stock (61.4%) of AI-powered apps, portion gaming has the smallest stock astatine 6.2%. Travel (12.3%) and Business (19.1%) are besides low-AI segments.

The much astonishing figures are astir AI apps’ quality to clasp their paying customers. AI apps underperform connected retention astatine some a monthly and yearly level, RevenueCat’s information shows.

Annual retention, a metric focused connected the app’s quality to clasp subscribers aft 12 months, was 21.1% for AI apps, compared with a higher 30.7% for non-AI apps. Monthly, AI apps saw 6.1% retention rates, versus 9.5% for non-AIs — a quality of 3.4 percent points.

The lone country wherever AI led connected retention was connected the play front, wherever AI apps had 2.5% retention rates compared with 1.7% for non-AI apps. It’s worthy noting that play subscriptions are not the astir fashionable enactment for AI apps.

Image Credits:RevenueCat

These metrics could beryllium influenced by the rapidly-changing authorities of AI technology, which could spot users hopping betwixt antithetic AI apps much quickly, arsenic they effort to find the 1 that has the astir existent exertion nether the hood.

AI vs non-AI apps by subscription program typeImage Credits:RevenueCat

As customers experimentation with a increasing fig of AI apps, they’re besides much apt to find that immoderate don’t conscionable their needs. The study notes that AI apps person 20% higher refund rates (4.2% vs.3.5% astatine the median) than non-AI apps do.

The precocious bound of refund rates for AI apps is besides higher (15.6% vs. 12.5%), suggesting there’s “greater volatility successful realized gross and deeper issues successful idiosyncratic value, experience, and semipermanent quality,” the study notes.

ScreenshotImage Credits:RevenueCat

There are immoderate benefits to being successful the AI-powered apps cohort, the information indicates.

RevenueCat recovered that AI apps person users from trials to paid customers 52% amended than non-AI apps (8.5% vs. 5.6% astatine the median), and AI apps monetize their downloads astir 20% amended than non-AI apps (2.4% to 2.0% astatine the median).

AI apps besides make 39% oregon higher monthly realized beingness worth (RLTV), a metric that measures the existent nett worth of an mean paying idiosyncratic implicit time. AI apps’ median connected this metric is $18.92 per month, compared with $13.59 for non-AI apps. AI apps besides prolong a 41% oregon higher RLTV connected an yearly basis, astatine $30.16 vs. $21.37, besides astatine the median.

The wide takeaway from the report’s findings is that AI tin thrust strong, aboriginal monetization, but these apps are struggling to prolong their worth with customers implicit time.