trial-to-paid conversion

Trial-to-paid conversion is the share of free-trial starters who go on to pay:

$\text{trial-to-paid} = \frac{\text{trials that ended in a first payment}}{\text{trials that have ended}}$

The denominator is the part people get wrong. Count only trials that have had time to finish (plus a day or two for the store to process the charge). Trials still running can't have converted yet; if you divide by all trials started in the period, the rate is too low, and it looks lowest for the most recent days, which can pass for a sudden drop. This is right-censoring: group trials by start date and read each group only once it has matured.

What rate to expect depends heavily on the trial type:

• Opt-out trials (a payment method is required, and the subscription renews automatically unless cancelled) convert a much larger share of starters, but fewer people start them. This is how app-store subscription trials usually work.
• Opt-in trials (no payment method; the user must decide to pay at the end) attract more starters but convert a smaller share.

So compare trial-to-paid only between like trials, and look at the product of the two steps too: paywall view → trial start → paid. A useful early signal on the app stores is auto-renew turned off during the trial: most people who will cancel do it in the first day or two, long before the trial ends. And the first payment isn't the finish line: check refunds and the first renewal before calling a trial change a success.