Trial Conversion Maturity and Completion Bounds Calculator

What the four counts mean

Use one frozen eligible cohort and one fixed horizon H after each trial start. A conversion is a qualifying post-introductory payment occurring no later than H. Introductory payments are excluded. “Mature” means the entire horizon is observable through a complete data watermark, even if the trial succeeded or cancelled earlier. This tool performs local arithmetic; it does not inspect your billing account or use audit credits.

Enter N, all eligible starts; M, units with the full horizon observed; C, conversions within H among those M; and E, already-observed conversions within H among the N−M immature units. One unit may appear in only one maturity group. Counts are integers from zero to one billion, with C ≤ M ≤ N and E ≤ N−M. Blank is missing, not zero.

Formulas and reconciliation

• Immature trials = N−M.
• Mature trials not converted within H = M−C.
• Immature outcomes unresolved = N−M−E.
• Maturity share = M/N.
• Mature conversion rate = C/M.
• All-starts observed conversion = (C+E)/N.
• Logical final lower bound = (C+E)/N.
• Logical final upper bound = (C+N−M)/N.
• Width of bounds = (N−M−E)/N.

Rates are undefined when their denominator is zero. The lower bound assumes none of the remaining immature trials convert within H; the upper assumes all do. These are logical completion bounds for this finite cohort, not confidence intervals, forecasts or best/worst profit scenarios. No probability distribution is assigned to the unresolved outcomes.

Reproduce the default example

Synthetic. N=1,000, M=600, C=180 and E=40. Four cells reconcile: 180 mature converted + 420 mature not converted + 40 immature converted + 360 immature unresolved = 1,000. Maturity is 60%; mature conversion is 30%; observed all-starts conversion is 22%; final bounds are 22% to 58%, a 36 percentage-point width.

If all 1,000 units become mature with 300 within-H conversions, enter N=1,000, M=1,000, C=300 and E=0. Both bounds then equal 30%. A later payment outside H does not change that fixed-horizon outcome. If you choose a longer horizon, create a new labelled measure rather than silently revising the old one.

When these bounds are not valid

The interval requires a permanently recorded conversion occurrence and complete historical coverage. If mature records are missing, reversed or reclassified, the lower and upper bounds may not contain the eventual corrected result. A net-of-refund outcome is a different measure and needs a refund observation window; do not remove refunds opportunistically from this occurrence count.

Early successes must remain immature until their horizon closes. Otherwise the denominator includes young successes while excluding equally young non-successes. The mature subset may also come from a different channel, market, offer or trial length. Its rate describes that subset, not a forecast for newer users or a comparison proving lift.

Make a reviewable export

The optional scope field stays local and is included in the JSON, along with raw counts, unrounded results and assumptions. Do not paste identities or secrets. Editing a value clears stale results; Reset example restores the documented fixture. Copy and download are explicit actions, not automatic uploads. Preserve the source query and offer version alongside the export.

For the source distinction between trial-start and customer-start cohorts, consult RevenueCat's trial conversion definition. The companion article shows a vendor-independent row classifier rather than claiming an exact platform match.