How to calculate trial-to-paid conversion from Stripe data

Trial-to-paid conversion is the share of trials that ended in the period and produced at least one paid invoice. Count trials that ended, not trials that started, leave out trials still running, and value the ones that did not convert as potential revenue, not as a loss. Here is the exact counting.

What counts as a converted trial?

A subscription with a trial start and end date, whose trial ended inside your window, and which has at least one paid invoice with an amount paid above zero created since the trial started. If it does, it converted. If not, it did not.

Stripe holds all of this: trial start, trial end and the invoice list. You do not need product data to count them. You do need product data to understand why they did not convert.

How do I calculate the rate?

1. Choose a window, such as the last 365 days.
2. List subscriptions whose trial ended in that window and before today.
3. Remove trials still running. They have no outcome yet.
4. Count how many have a paid invoice since trial start.
5. Divide converted by ended.

Leaving out running trials is the same logic as leaving out failed invoices still in retry. A trial that ends tomorrow cannot be called a failure today, and including it pushes the rate down for no reason.

What does the synthetic sample show?

In the fictional company in my sample report, 426 trials ended in the window and 177 never paid, a conversion of 58.5%. Seventeen more were still running and were excluded. I generated this data, so it tells you nothing about your account, and I would not compare your rate to it. I do not have a trustworthy benchmark, and I would rather say so than quote one.

How do I put dollars on the trials that did not convert?

Take the monthly price of the plan each one trialled, add them up, and divide by the months in the window. For the sample that is about $989 a month. The number is a ceiling: it is what those trials would have been worth if every one had paid, and not every one would have.

That is why I keep it out of the headline and show it as a labelled potential. For the full argument, read unconverted trials: what are they worth.

What do I do when conversion is low?

First find out why. The mechanism I work from: trials that reach a first moment of value tend to convert, and those that never do rarely do. Stripe cannot show activation. You need product events from your own app, such as whether someone created a first project. That is the main gap between a billing export and what you need to act.

Billing itself causes two of the failures. Some people forget they started a trial and are surprised by the first charge. And some have a card that fails on that first charge, which is a payment recovery problem. For that, see how to recover failed payments.

What can I try first?

• Nudge unactivated trials early, toward the one step that gets them to value. The last day is too late for someone who never started.
• Remind people about the card and the first charge before the trial ends, with the date and the amount.
• Review the trial design only afterwards: its length, card up front or not, what is included. Those are product decisions that billing data cannot settle.

I would try the first two before the third. They are cheaper and easier to reverse.

How do I check a trial change worked?

With a split. Send the nudge to a random half, or to everyone except a holdout, and compare how many in each group paid. Keep the split fixed. Use net revenue per customer, after margin and after the cost of any offer, since a trial that converts only because you gave a discount may not be a win.

With fewer than about 30 conversions in a group, you have an early signal and nothing more. See what a holdout group is.

What are the traps?

• Currency. Trials on plans in other currencies are excluded unless you state a rate.
• Plan changes during a trial. The value used is the plan on the subscription, which may differ from what the person first picked.
• Usage-based plans. No fixed price exists, so the figure is an estimate from the last paid invoice.
• Conversions outside Stripe. If you convert people in another system, Stripe sees a trial with no payment.

How often should I recompute it?

Monthly is enough for most accounts. Use the same window length and the same rules each time, and record the date if you change one. Trials ending in a given month take time to show their outcome, so the latest month will always look a little worse until its stragglers pay. Compare full months with full months, and treat the current month as a preview.

Should I start with trials?

Compare in dollars and discount the trial figure because it is potential. The payment fix and the cancel flow are realised numbers, so unless the trial gap is much larger, start there. The Revenue Leak Report lays all four leaks side by side for $490, within 48 hours of my having a restricted read-only Stripe key, so you can decide from your own data.