Free trial vs no free trial (and how long a trial should be)

There is no universal answer to free trial vs no free trial. A trial helps when people need to use the product with their own work before they believe it does what the page says. Charging up front, often with a money-back guarantee, works when the value is clear before use or when each trial user costs real money to serve. Removing a trial cuts the number of people who start, because paying is a bigger step than trying. Whether it cuts revenue depends on how many of those trial users would ever have paid.

So test it. Split new visitors between the two versions and judge on revenue per visitor, counted inside a window that covers the longest trial plus one billing cycle. For trial length, the rule of thumb is long enough to reach first value and no longer, and the data below shows why longer trials look better than they are.

The options

What the data says

The best public dataset is from mobile apps, not SaaS, so treat it as a direction rather than a benchmark for your business. RevenueCat's State of Subscription Apps 2026, covering over 115,000 apps, reports:

• Apps with a hard paywall (pay or trial before using) had a median download-to-paid rate of 10.7% by day 35, against 2.1% for freemium apps.
• Trials of 17 to 32 days had a median trial-to-paid rate of 42.5%, against 25.5% for trials of 4 days or less.
• 55.4% of all cancellations of 3-day trials happened on the day the trial started.

The trial length numbers need care. RevenueCat's own write-up on trial length calls the pattern correlational, since someone still in a trial on day 25 is more likely to pay than someone who left on day two. Apps that choose long trials may also differ from apps that choose short ones. The data says long trials can work. It does not say lengthening your trial will add 17 points of conversion.

The day-one cancellation figure says something more useful. Many people decide in their first session. Whatever the trial length, the first session carries a lot of the weight, which is why onboarding tests often pay off alongside trial tests.

How to test trial vs no trial

1. Randomize new visitors only. People who have already seen one offer shouldn't see another.
2. Change only the trial. Keep the price, plans and page the same.
3. Fix the measurement window before launch. It starts at each visitor's first visit and covers the longest trial plus one billing cycle. For a 14-day trial on a monthly plan, that is about 45 days.
4. Count revenue per visitor inside the window, minus refunds and chargebacks.
5. Plan the calendar. Enrollment takes as long as your traffic needs, then you wait for the window to close for the last visitor. The test duration calculator gives the enrollment part.

A worked example

This is a made-up example. A product costs $40 a month. Each version gets 10,000 new visitors, and revenue is counted for 45 days from each visitor's first visit.

Here is the maths. A collects (270 + 216) × $40 = $19,440. B collects (202 + 172) × $40 = $14,960. C collects (248 + 193) × $40 = $17,640. Divide each by 10,000 visitors.

In this example, the 14-day trial earns the most per visitor. Now look at what happens with the wrong window. Ten days after each visit, A has collected nothing, because no trial has ended yet, while B has already taken $8,800 less early refunds. A team that reads the test early would kill the version that wins. That is the most common mistake in trial tests.

Also note what the example does not show. Trial users who convert may churn at a different rate from people who paid up front. Look at revenue per visitor again at 90 days before you treat the result as settled.

How long a free trial should be

Start from the product, not from a benchmark:

• Find the moment a new user first gets real value, such as the first invoice paid or the first report shared.
• Measure how long a typical successful user takes to get there. The trial should cover that with room to spare.
• Match the product's natural cycle. A daily habit app may need 3 to 7 days. A monthly workflow may need 14 to 30.
• Count the cost to serve. If each trial user costs you money, such as compute for an AI product, a long trial full of people who never pay gets expensive.

Then test two lengths against each other, such as 7 vs 14 days, and judge on revenue per visitor in a window that covers the longer trial plus one billing cycle.

Rules to follow

Trials that turn into paid subscriptions have their own rules. This is general information, not legal advice.

• In California, the automatic renewal law as amended from July 2025 covers free-to-paid conversions. For a free trial longer than 31 days, the business must send a notice 3 to 21 days before it converts, per Cooley's summary.
• In the UK, the subscription contracts rules in the Digital Markets, Competition and Consumers Act are due to start in January 2027 and include a 14-day cooling-off period after the first payment following a free or discounted period, according to TLT's briefing.
• Google Play's subscriptions policy requires apps to tell users how and when a trial converts, what it will cost and how to cancel.
• Apple's subscription guidance asks apps to show how long the trial lasts and the price billed after it.

A trial variant that hides the conversion date to win more starts is not a valid test. It will lose on refunds and complaints, and it may break these rules.

Tests to try, and what to measure

1. Trial vs no trial with a money-back guarantee. Measure revenue per visitor at 45 and 90 days and the refund rate.
2. Card required vs no card. Measure revenue per visitor after the trial window, trial starts, refunds and chargebacks.
3. 7-day vs 14-day trial. Measure revenue per visitor in a window covering the 14-day trial plus a billing cycle.
4. A reminder email 3 days before the trial ends vs none. Measure revenue per visitor and refund requests.
5. A $1 paid trial vs a free trial. Measure revenue per visitor and trial starts.
6. Reverse trial vs plain freemium. Measure revenue per visitor at 60 days and the share of free users who upgrade.

Outtest can run trial vs no trial and trial length tests and judges them on revenue per visitor from your payment tool. Because a trial length change touches your app, it goes through a GitHub pull request that you approve. If your traffic is small, read A/B testing with low traffic before you start, since trial tests take longer than most.