How Long Should a Free Trial Be in 2026?
5 min
A trial of 17 to 32 days converts at a 42.5% median. A trial of four days or less converts at 25.5%. The gap is 17 percentage points (42.5 minus 25.5). Those medians are from RevenueCat's State of Subscription Apps 2026. RevenueCat also writes that nearly half of apps now use a trial of four days or less.
A is the middle app in the group. The top quarter of the long-trial group sits above 59.4%. The top quarter of the short-trial group sits above 38.5%. A trial of 5 to 9 days sits between them, at 37.4%.
0.0%
17–32 day trials
0.0%
5–9 day trials
0.0%
4 days or less
These three numbers mix every plan. A weekly plan and an annual plan do not convert the same way. The picker below splits them.
Does the longer trial keep paying?
Conversion here means the share of trial starts that become a paid subscription. Renewal means the share of those payers who pay a second time. A high conversion with a weak renewal is one payment.
The plan split comes from a RevenueCat follow-up on more than 17,000 apps. The window is trial starts from August 2025 through July 2026. Each rate is a median of apps, not one pool of users.
Plan
Trial length
46.6%
Conversion
72.0%
Renewal
Median of apps with more than 100 trials in the cell. Groups under 20 apps were left out. Trials longer than 32 days are not in the conversion figures. August 2025 to July 2026.
On a weekly plan, 5 to 9 days converts 24.3%. Of those payers, 65.9% renew. Four days or less converts 22.3%. Of those payers, 57.9% renew. Past nine days, RevenueCat says the weekly sample is too small to compare.
On a monthly plan, 10 to 16 days is the high-conversion cell: 46.6%. Of those payers, 72.0% renew. Four days or less converts 39.6%. Of those payers, 54.2% renew. The 17 to 32 day group renews highest, at 77.5%. Conversion there slips to 43.7%.
On an annual plan, 17 to 32 days converts 44.6%. Of those payers, 47.5% renew. Four days or less converts 24%. Of those payers, 18.3% renew. That short annual trial also renews worse than no trial, which renews at 26.6%. RevenueCat did not publish conversion for the annual middle lengths. The picker leaves those blank.
A longer trial and a more patient buyer sit in the same cell. RevenueCat notes that someone still in the trial on day 25 was already more likely to pay than someone who left on day 2. Test the longer length. Do not paste 42.5% onto a paywall by editing the day count.
When do people cancel?
The report counts cancellations, then asks which day they fall on. It does not say that 55% of all 3-day trials are cancelled. It says that, of the 3-day trials that are cancelled, 55.4% of those cancellations happen on day 0. 84% happen on day 0 or day 1.
The same day-0 share is 39.8% for a 7-day trial (64% by the end of day 1), 35.7% for 14 days, and 31.1% for 30 days. A short trial does not create a decision at the deadline. For most cancelled 3-day trials, the decision was the first session.
When is the long trial the wrong default?
Each extra free day costs money when the session itself has a bill. In the AI subset of the same study, a monthly trial of 10 to 16 days converts at 38.5%. Of those payers, 64.2% renew. A monthly trial of 17 to 32 days converts at 31.8%. Renewal stays at 64.1%. In that subset the longest monthly trial is not the stronger cell.
What should you set first?
Pick the length the table actually measures for the plan you sell.
- Weekly: test 5 to 9 days against 4 or fewer. Longer weekly trials are not in the published sample.
- Monthly: start the test at 10 to 16 days. Try 17 to 32 days only if people still open the app in week three.
- Annual: the short trial is the weak cell. Test 17 to 32 days before you copy a 3-day annual offer.
Then read your own conversion and your own renewal. These medians are a comparison, not a score you have earned.
I set the trial from the plan and from what a free session costs. If you want a second look at a paywall before you ship it, let's talk.
Sources
Building something?
Work with me