TractionScout
← Back to the blog
Honest growth metrics

Trial-to-paid conversion: measure it by cohort, not by ratio

Count trials by the day they started, give each one its full time, and count only real payments

7 Nov 2026 · 5 min read

Trial-to-paid conversion rate is the share of trials that started in a period and became a paid subscription within the trial plus a short allowance. Count trials by the day they started, wait until each one has had its full time, and count only settled payments. Dividing this month's new customers by this month's new trials looks similar but measures nothing.

Why the monthly ratio is wrong

The quick version takes paying customers who converted in March and divides them by trials that started in March. With a 14-day trial, many of March's conversions started their trial in February, and many of March's trials have not ended yet. The two numbers describe different people.

It also swings for the wrong reasons. A launch in late March floods the denominator with trials that cannot have converted yet, so the rate collapses in the best month you have had. The next month it jumps, because the conversions arrive without their trials.

Count by trial start, not by payment date

A cohort is the group of trials that started in the same period, followed forward. Each trial belongs to the day it started, forever, and its conversion is credited back to that cohort whenever it happens. The rate for a cohort is converted trials divided by trials, for the same people.

Daily or weekly cohorts work well for small products: they show change quickly without mixing launches and quiet weeks.

Give every trial its full time

A trial can only count as converted or not once it has had the chance: its trial length plus a short allowance for the first payment to settle, such as seven days after the trial ends. Until then it is pending, not a failure.

So the rate you report should use mature trials only, those whose window has passed. Show pending trials separately, so a growing pile of recent trials reads as "too early" rather than "conversion collapsed".

A worked example

Fictional weekly cohorts for a 14-day trial, read on November 30, with a 7-day allowance (21 days in total):

Trials startedTrialsConverted so farWindow passed?Rate
Week of Oct 26409Yes22.5%
Week of Nov 25211Yes21.2%
Week of Nov 9618PartlyPending
Week of Nov 16120 (launch)3NoPending

The launch week has three times the trials and almost no conversions yet. A monthly ratio would call November a disaster; the cohorts say conversion is steady at about 22% and the launch cohort simply has not finished.

What counts as a conversion

A trial converts when the customer pays: a positive, settled payment for the recurring plan. Several things look like conversion and are not:

  • A saved card or a payment method on file. It is permission to charge, not a charge.
  • A zero-amount invoice or a created subscription. Billing systems create both at trial start.
  • Someone else's payment. The conversion must belong to the same customer who started the trial.

A customer who paid and later cancelled still converted; cancellation is a separate measure. And if you cannot link payments to trials at all, the honest answer is "not measured", not 0%.

Card-required and card-free trials are different products

Asking for a card at sign-up usually means fewer trials and a much higher share who pay; a card-free trial brings more trials and a lower rate. Neither number is better on its own. Compare paid customers per visitor if you are deciding between them, and never compare a card-required rate with someone else's card-free benchmark.

What moves the rate

Most trials are lost before the billing date: the user never reached the product's first useful result. Look at activation during the trial first (see Activation rate: find where new sign-ups get stuck). Then check the end of the trial: a clear reminder a few days before it ends, and a pricing page that says what the paid plan adds.

How TractionScout measures it

TractionScout Performance for a sample product: the Converted step with sign-ups charted daily, and trial starts, successful payments and app downloads beside it

Performance in the TractionScout demo workspace on sample data: the Converted step leads with sign-ups, with trial starts and successful payments beside them.

TractionScout puts trials, trial conversions and trial-to-paid rate on each project's board, on the Converted step, with trends and a comparison with the previous period. They are cohorts by trial start day: each period counts the trials that started in it, how many have converted so far, and, for the rate, the share of mature trials that converted.

When Stripe is connected with confirmed billing accounts, a trial converts on a positive, settled recurring invoice by the trial's end plus seven days. Otherwise the source that started the most trials, such as RevenueCat or the TractionScout SDK, is the authority, and a conversion must belong to the same customer. When lifecycle history is incomplete or conversions cannot be linked to their trials, the figure is withheld instead of shown as zero. See Smart KPIs.

Questions

What is a good trial-to-paid conversion rate?

It depends on whether you ask for a card, your price and your audience, so outside benchmarks rarely transfer. Track your own mature cohorts and improve on them.

What if I change the trial length?

Trials under the new length are a new series. Compare them with each other, and give them their new, longer or shorter window before reading a rate.

My trials have no fixed end date. How long should I wait?

Use the longest normal trial plus the allowance. TractionScout uses 37 days for trials without a known end: a 30-day trial plus seven days.

Empfohlen

See trial-to-paid by cohort for your own product

Trials, trial conversions and trial-to-paid on each product's board, counted from settled payments.