Two-stage Retention Curve Calculator

Model a retention curve with different churn rates for early months vs steady-state, and estimate expected value over time.

Many products have high early churn (activation/onboarding) and lower steady-state churn later. A two-stage model can match reality better than constant churn.

This calculator builds a simple retention curve with separate early and steady-state churn rates and estimates expected revenue and gross profit per original customer.

Prefer an explanation- Read the guide.
 
%
 
 
%
 
$
 
%
 
Tip: you can type commas (e.g., 10,000).

Example

Using the default inputs, the result is:
75.9%
Early monthly churn
6%
Early phase months
3
Steady-state monthly churn
1%
ARPA (monthly)
$800
Gross margin
80%
Months to model
36

How to calculate

  1. Set an early churn rate and how many months it applies (e.g., months 1-3).
  2. Set a steady-state churn rate for later months.
  3. Enter ARPA and gross margin to convert retention into expected value.

Formula

Retention(m) = (1 - churn_early)^(min(m, earlyMonths)) x (1 - churn_steady)^(max(0, m - earlyMonths))
  • Logo churn is modeled in two phases (early vs steady-state).
  • ARPA and gross margin are constant over the horizon.
  • Outputs are per original customer/account (expected value).

FAQ

When should I use two-stage churn-
When you observe a clear activation/onboarding drop early and much lower churn later. Two-stage models let you stress-test the impact of improving early retention vs improving steady-state retention.
Does this replace cohort analysis-
No. It's a planning shortcut. Real cohort curves (by segment) are the gold standard for understanding retention dynamics and forecasting LTV.

Common mistakes

  • Using churn rates from blended segments (plan/channel).
  • Treating this as a substitute for real cohort curves; use it for planning and sensitivity.
  • Ignoring expansion (revenue retention) when it's a major driver of value.

Quick checks

  • Keep time units consistent (monthly vs annual) across inputs and outputs.
  • Segment by cohort/channel/plan before trusting a blended average.
  • Use the related guide to avoid common definition and denominator mismatches.