4 min read · Pleroma Commerce · 3 sources
In short
AppsFlyer data on 2,900 subscription apps shows that trial volume and paid conversion move in opposite directions: gaming starts the most trials and converts 19% of them, education converts 42%. In concentrated categories the top five apps take over 90% of ad spend, and only a different product angle moves share. Pure subscription is giving way to hybrid models. For a subscription brand on Shopify, the lesson is to measure second and third payments by source, not sign-ups.
In this article
AppsFlyer's new subscription report covers 2,900 subscription apps, 1.7 billion paid installs and 2.1 billion dollars of acquisition spend between October 2024 and February 2026[1]. The headline numbers are growth: acquisition spend up 24% year on year, Android paid installs up 57%. The more useful numbers are underneath, and they describe a problem every subscription business has, whether it sells an app or a box of coffee: getting someone to start is easy, getting them to pay again is the business.
One caveat before the lessons. This is mobile app data, dominated by streaming, short video, dating and fitness apps. A Shopify brand selling pet food or skincare on subscription is not in the sample. The mechanics transfer. The benchmarks do not, so treat the percentages as direction, not targets.
A trial is a question, not an answer
Gaming apps pull 12.2% of installs into a free trial, the highest rate in the dataset, and convert only 19% of those trials into paying subscribers, the lowest[1]. Education converts 42% of trials and Lifestyle 41%, even though Lifestyle gets the fewest people to start one. Across 16,000 apps, Adapty puts the average trial-to-paid rate at about 26% and finds that nine in ten trials start on the day of install[2]. People decide to try almost immediately. Whether they decide to pay is a different question, answered later.
The e-commerce equivalent is the first box for one euro, or the 50% off first order. Both produce a sign-up number that looks like growth. Neither says anything about the second order. A subscription brand that reports trial starts instead of second payments is measuring how cheap its offer is, not how good its product is.
Budget does not beat a head start
In short drama and streaming apps, the top five players take more than 90% of acquisition spend, and that share has not moved[1]. In health and fitness it jumped from 54% to 73% in one year: spend rose 87% on iOS, and almost all of it went to a handful of apps that could afford the auction. RevenueCat's 2026 report, built on more than 115,000 apps, shows the same divide from the revenue side, with the top quarter growing recurring revenue by 80% while the rest fall behind[3].
The one category moving the other way is photo and video, where the top five share fell from 64% to 45% as AI-based tools arrived with something the incumbents did not have[1]. The lesson for a smaller subscription brand is uncomfortable but clear. In a concentrated category, more ad budget buys a seat in an auction the leaders have already won. A different product angle is the only thing that moves share.
Pure subscription is not the only model
Monetisation is splitting by category. Streaming apps are simplifying: the share making money from subscriptions only rose from 53% to 62%[1]. Short drama apps are going the other way, adding ads to earn something from users who will not subscribe, with that model rising from under 1% to 7.4% of apps. Generative AI apps are adding one-off purchases for people who will not commit to recurring billing, and subscription-only apps in that category fell from 65% to 60%.
For a physical subscription business the translation is direct. Let subscribers add one-off products to their next box. Let one-off buyers turn into subscribers after the second order, not before the first. Offer pause and skip before cancel. A customer who pauses is still a customer. A customer forced to choose between subscribing and leaving often leaves.
Ask why, not which
AppsFlyer analysed more than 4,000 questions subscription marketers asked its AI assistant. Retrieval questions outnumber diagnostic ones by about ten to one: which campaign had the best return, rather than why return dropped[1]. Within the questions about return on ad spend, only 6% were diagnostic. The exception was business apps, where questions about retention and cohorts ran at about four times the average, because the business model forces lifecycle thinking.
Every subscription business should be forced into it. The question that matters is not which channel brought the most sign-ups this month. It is which channel brought the subscribers who were still paying in month three.
At Pleroma we measure subscription programmes on second and third payments by acquisition source, before anyone scales the budget.
A trial is not a customer. The second payment is.
Pleroma Commerce
Measure the second payment
A five-day audit of the store and every channel, including subscription retention by acquisition source.
See the store and channel audit →Sources
- AppsFlyer, "The State of Subscriptions for Marketers, 2026 Edition", March 2026. appsflyer.com
- Adapty, "State of in-app subscriptions 2026". adapty.io
- RevenueCat, "State of Subscription Apps 2026". revenuecat.com