Annual trials of 30 days converted 44.6%, four-day trials 24%

Most founders set their trial length once and leave it. A dataset published by RevenueCat, the subscription billing firm, gives them a reason to look again.

The study covers more than 17,000 mobile apps over a year, from August 2025 to July 2026. RevenueCat says it handles subscription management for around 60% of mobile subscription apps, which is where the sample comes from. The findings were summarised by Jason Lemkin of SaaStr, whose fund invested in RevenueCat in 2018.

On annual plans, conversion rose at every step up in trial length, according to the study. Trials of four days or less converted 24% of users. Five to nine days converted 33%, 10 to 16 days converted 43%, and 17 to 32 days converted 44.6%.

First renewal moved further, from 18.3% on the shortest trials to 47.5% on the longest. Combine the two and the share of trial starters who paid and were still paying a year later went from 3.5% to 18.5%.

Monthly plans behave differently. Conversion peaked at 46.6% on trials of 10 to 16 days, then fell to 43.7% on trials of 17 to 32 days, while first renewal kept climbing to 77.5%. For a self-serve monthly product, the familiar 14 days sits inside the best band.

Dropping the trial altogether carries a measurable cost on monthly plans. Buyers who paid with no trial renewed at 49.5%, against 77.5% for those who came through a trial of 17 to 32 days.

Annual is the exception. No-trial annual buyers renewed at 26.6%, ahead of those who came through trials of four days or less, who renewed at 18.3%. On the study's numbers, a short trial paired with an annual push produced the weakest cohort in the dataset.

AI products hit a ceiling earlier. On monthly plans, trials of five to nine days and 10 to 16 days converted at 38.2% and 38.5%. Past 16 days, conversion fell to 31.8% and renewal stayed flat at 64.1%, so two extra weeks of free inference bought fewer customers and no extra retention.

Lemkin's advice for AI founders is to cap time-based trials at about two weeks and give annual buyers more room through usage limits — credits, runs or seats — rather than more days on the clock.

Geography matters too. North America and Western Europe converted better at every longer annual trial length. In the Middle East and Africa, monthly conversion was 38.3% at five to nine days and 27.4% at 17 to 32 days.

RevenueCat flags a caveat in its own post: the results are correlational. A user still active on day 25 was more likely to buy than one who left on day two, whatever the trial length, so some of the lift comes from who stays rather than from the length itself.

The practical move is narrow and cheap. Run 14 days against 30 days on your annual offer for a quarter, and measure conversion and 12-month renewal together rather than one at a time.

The sample is mobile, and skews consumer: fitness apps, photo editors, games, language learning. Whether an annual software contract behaves like an annual app subscription is the question a founder has to answer on their own funnel.