Your trials
Your trial and plan
Use one row per trial and dates like 2026-03-14 (a timestamp is fine, only the date is read). A trial still running as of the date above is left out, not counted as a failure. Leaving the last box at 0 gives the worst case: every conversion after the shorter cutoff is lost.
Upload or paste your trials, then click Analyse.
Result
Conversions by day of the trial
What a shorter trial would have cost
Counts conversions that came after the shorter cutoff, from your finished trials. The range is the middle 80% of 300 seeded resamples of those trials. It cannot see what a shorter trial does to sign-ups, and it says nothing about trials longer than the one you ran.
| Trial of | Conversions after the cutoff | 80% range | Lost per 100 trials | Value lost per 100 trials |
|---|
Back-test
How it is calculated
Finished trials only: a trial counts once start + trial length is on or before the as-of date. Conversion day is first payment minus start. The share of conversions after day T is counted directly; the conversion rate under a T-day trial is your rate minus those late conversions, less the share you say would have paid sooner anyway. Ranges come from 300 seeded resamples of the finished trials. The back-test compares the share of conversions after the halfway day in the first 75% of trials (by start date) with the last 25%, and calls the timing unstable if they differ by more than 15 points.
Not modelled: what a shorter or longer trial does to sign-ups, to who starts a trial, to activation or to later churn; reminder emails and discounts; card-upfront versus no-card trials; and anything about trials longer than the one you ran. Users who pay and cancel are counted as converted. Unverified: tested on simulated trials with known timing, not on a published dataset. Changing the trial length is an experiment; this shows what your history allows you to expect, not a result.
Common questions
What is a good trial length?
There is no universal answer, because it depends on how long your product takes to show its value. The useful evidence is your own: the day each trial user first paid. If most of your conversions arrive in the first few days, a long trial mostly adds waiting. If they bunch at the end, people may be waiting for the deadline, and shortening the trial would pull them forward rather than lose them. This tool shows which pattern your trials have.
How do I see when my trial users convert?
Export the date each trial started and the date each user first paid, leave the payment blank for those who did not, and paste the two columns here. The tool draws conversions per day and the running share, and reports the median and the 90th percentile day.
What would shortening my trial cost?
The tool counts the conversions that happened after the shorter cutoff. That is the worst case, because it assumes none of those people would have converted sooner. You can set the share you believe would convert anyway, and the loss scales down with it. It is a range from your own history, not a forecast, and it cannot tell you what a shorter trial does to sign-ups.
Can it tell me whether a longer trial would convert more?
No. Your history has no trial longer than the one you ran, so nothing in it says what days beyond that would add. The tool only looks inside the length you offered, and says so. To learn about a longer trial you would have to run one.
What is the deadline effect in trial conversion?
Many users pay on the last day or two of a trial because the deadline is what makes them decide. The tool flags when at least 30% of your conversions fall in the last two days and that is well above a flat spread. In that case the days before the deadline are not evidence of low interest, and a shorter trial could move the pile-up earlier.
How many trials do I need?
At least 30 finished trials and at least 8 conversions. A trial only counts once its whole window has passed, so a trial that is still running is left out rather than counted as a failure. With under 25 conversions the tool calls the result weak and asks you to read the ranges.
Why does it say my timing is unstable?
It fits on the first 75% of your trials by start date and checks the last 25%. If the share of conversions arriving after the halfway day differs by more than fifteen points between the two, your conversion timing has moved, for example after a pricing or onboarding change, and an old average would mislead you.
Is my customer data uploaded anywhere?
No. The dates are read and analysed in this browser tab. Nothing is sent to a server, there are no accounts or cookies, and the tool does not store your trials. It works offline once the page has loaded.