SaaS Cohort Retention Calculator

Retention from month 3 to month 12 determines LTV — not month 1. Measure it, or you'll pay CAC twice for the same customer.

  • Instant result
  • No sign-up
  • Visible assumptions
  • Deterministic calculation

In 30 seconds: Simulate the retention curve month by month by cohort and separate real growth from revenue that only comes from CAC burned on new logos. Deterministic calculation with auditable formulas. The result is indicative — adjust the assumptions to reflect your real operation.

Methodology

Active users in month m = Cohort size × (1 − monthly churn)^m

Retention (%) in month m = (1 − churn)^m × 100

Monthly revenue in month m = Active users × ARPU

Cumulative revenue = Σ Monthly revenue from month 1 to horizon

Variables

Cohort Size
Number of users or accounts that entered in the same period.
Monthly ARPU
Average monthly revenue per user.
Monthly Churn
Percentage of users who leave each month.
Horizon (months)
How many months forward to project the cohort.

Practical example

A cohort of 500 users enters in January with $299 ARPU and 6% monthly churn.

Month 3 — retention: (1 − 0.06)^3 = 83.1% → 415 active users.

Month 6 — retention: 69.0% → 345 active users.

Month 12 — retention: 47.6% → 238 active users.

Cumulative 12-month revenue ≈ $1,055,000 — useful for comparing against the initial investment in acquiring that cohort.

Interpretation

Month-3 (M3) retention is the most reliable early indicator of product health: cohorts that survive the first 90 days tend to stick around much longer.

The exponential curve is a floor: reality is often better due to late activation or reactivations, or worse due to shocks (pricing changes, outages).

Compare cohorts month over month to spot product improvements or regressions. If March's cohort retains better at 3 months than February's, something shifted in your favor.

If your M12 retention is below 30%, your business depends heavily on acquiring new customers to grow — retention is a priority lever.

Assumptions and limitations

  • Assumes constant churn (pure exponential decay). In practice, churn is often higher in the first months and then stabilizes.
  • Assumes constant ARPU: doesn't account for upgrades, downgrades or retention discounts.
  • Does not model reactivations (users who return after canceling).
  • Treats the cohort as homogeneous — if there are segments with very different churn, model each separately.

When to use this calculator

  • To project future revenue from an acquisition campaign before spending it.

  • When comparing acquisition channels: an organic cohort usually retains better than a paid one.

  • To set retention goals by milestone (M3, M6, M12) and give the product team a quantitative north star.

  • Before changing pricing: simulate with higher projected churn to estimate how much current MRR is at risk.

  • In post-incident analysis: if a product failure pushed monthly churn from 5% to 9% in one month, this calculator quantifies the damage in 12-month cumulative revenue.

Common mistakes

  • Assuming a single cohort represents all. Cohorts from different months can behave very differently — always compare at least 3.

  • Taking last month's churn as a constant monthly churn rate. Better to use the average monthly churn of the last 3-6 months.

  • Ignoring cumulative revenue and looking only at retention: a cohort with high churn but high ARPU can be more profitable than one with low churn and low ARPU.

  • Confusing user retention with revenue retention. If you have downgrades, users stay but revenue falls.

Industry use cases

B2B SaaS

Healthy cohorts retain 85-90% at M3, 75-85% at M6 and 65-75% at M12. Cohorts with M12 retention < 50% indicate a serious fit issue.

Consumer apps

Typical retention is much lower: 20-30% at M3 and 5-15% at M12. The strategy is to maximize revenue per user in the first 90 days or build habit.

Education platforms

Highly seasonal cohort usage: back-to-school cohorts retain better than mid-year ones. Model each seasonal cohort separately.

Marketplaces

Supply- vs demand-side retention diverges: sellers who make a successful sale in M1 retain at twice the rate. Useful to model two separate cohorts.

Methodology and assumptions

How results are calculated, what we assume when modeling, and where the method loses precision.

Formula

Retention(t) = Active customers in month t ÷ Customers in month 0 · GRR = 1 − Revenue churn

Assumptions

  • Non-overlapping monthly cohorts.
  • Active customer defined by usage or payment, depending on the metric you enter.
  • GRR / NRR is computed on revenue, not on logo count.

Applicability limits

  • Cohorts smaller than 50 customers produce noisy curves — interpret with margin.
  • Survivorship bias understates churn when data comes filtered from a CRM.
  • Exogenous events (pricing changes, launches) distort cross-cohort comparisons.

Sources

You projected your cohort. Add it to the cash flow simulator to see the impact on runway and ending cash. Advanced Cash Flow Simulator

Want to go beyond the quick calculation?

The advanced simulators model complete scenarios — 12-month cash flow, pricing with sensitivity analysis, credit risk, delivery routes — with your own data and no sign-up.

Explore the simulators

Calculator guide

What it calculates and who it is for

This calculator projects how a customer cohort fades month by month and how much revenue it leaves along the way. From cohort size, ARPU, churn and a horizon, it builds the retention curve, the cohort triangle, the bottom-up cumulative LTV and the half-life (the month the cohort drops to 50%).

It is aimed at product, growth and finance teams in SaaS and subscriptions who want to go beyond average churn: the same churn rate produces very different economics depending on whether losses concentrate early (smile curve) or spread evenly, and this engine models both.

Inputs

Cohort size
Customers who started together in month zero: the sign-ups of a single acquisition month.
Monthly ARPU
Average revenue per active customer per month, as actually charged to that cohort.
Monthly churn (%)
The base cancellation rate. In the constant model it applies identically each month; in the other models it anchors the curve's calibration.
Horizon (months)
How many months to project: 24 or 36 reveal the cohort's tail, where real LTV is decided.
Curve model
Constant (identical churn each month), Weibull (high early churn that stabilizes — the typical SaaS pattern, used by default) or two-phase (early and stable churn with an explicit crossover month).
Gross margin, expansion, CAC (optional)
Refinements: margin for LTV (70% by default), monthly expansion among survivors for NRR, and CAC for the cohort's LTV:CAC ratio and payback.

Results you get

Retention curve
The share of the cohort still active each month, with the associated active users and monthly revenue.
Half-life
The month retention crosses 50%, interpolated between the two straddling months. It compresses the curve into one comparable number.
Bottom-up LTV
The month-by-month sum of ARPU × retention × margin: the real value generated per initial customer within the horizon, with no extrapolation.
Cohort cumulative revenue
Everything the cohort billed over the horizon, and its gross-margin equivalent.
Cohort triangle
The classic retention-by-cohort-age table, ready to compare against your real data.
GDR and NRR
Gross dollar retention (no expansion) and net revenue retention (with the monthly expansion you enter), month by month.

Methodology and assumptions

Constant model: Retention(m) = (1 − churn)^m

Weibull model: Retention(m) = e^(−(m/η)^β), with β = 0.7 by default

Two-phase model: early churn until the crossover month, stable churn afterward

Bottom-up LTV = Σ ARPU × Retention(m) × Gross margin, for m = 1 … horizon

NRR(m) = Retention(m) × (1 + cumulative expansion), with gradual expansion ramp-in

In the default Weibull model, β below 1 produces the 'smile curve': the cancellation rate is high in the first months and decreases among those who stay. The scale parameter η is auto-calibrated from your churn so that 12-month retention matches the constant model — making both models comparable from the same input.

LTV is built by summing each projected month's real margin within the horizon, instead of the ARPU ÷ churn approximation, which overstates value when early churn is front-loaded.

Worked example

Hypothetical example for illustration. The numbers reproduce exactly when entered into the calculator on this page.

Worked example: a 500-customer cohort with $400 ARPU, 5% base monthly churn, a 24-month horizon and the default Weibull model (70% gross margin).

Month 2: 419 customers remain active (83.9% retention) — the early drop concentrates onboarding losses.

Month 7: 328 active (65.6%). Month 13: 261 active (52.2%).

Half-life: 14.2 months — where the cohort crosses 50%.

Month 24: 184 customers still active (36.8%). Cohort cumulative revenue: $2,703,142; cumulative gross margin: $1,892,199.

Bottom-up LTV: $1,892,199 ÷ 500 = $3,784 per initial customer within the 24-month horizon.

Reading: although the 'average' churn is 5%, the Weibull curve shows much faster early losses and slower late ones — half the cohort left before month 15, yet a third is still present at two years.

How to interpret the result

The curve's shape matters more than average churn: two businesses with the same 12-month churn have different LTVs if one loses its customers in the first quarter and the other loses them spread out. Early decay is attacked with onboarding; late decay, with product and customer success.

Use half-life to compare cohorts against each other: if recent cohorts have shorter half-lives than older ones, acquisition quality or onboarding degraded, even if the business's aggregate churn doesn't show it yet.

The horizon-bounded bottom-up LTV is deliberately conservative: it assumes no revenue beyond what is projected. Comparing it against your CAC gives a more honest payback reading than closed-form LTV.

Limitations and when not to use it

  • It is a parametric projection, not an analysis of your data: your cohorts' real curve may differ from the chosen model. Use it to structure hypotheses and contrast with your real triangle as soon as you have history.
  • It assumes constant ARPU among survivors except for the expansion parameter: plan changes and repricing are out of scope.
  • It models one homogeneous cohort: mixes of segments with different churn require one projection per segment.
  • A frequent mistake is entering the whole business's average churn, which blends mature and recent cohorts: to project a new cohort use the churn observed in young cohorts.
  • Do not use it as a formal financial forecast or for valuation: it is an operational retention-analysis tool.

From theory to calculation

The calculator on this page runs with your numbers — no forms, no login. Scroll up and try it.

Try the calculator

Frequently asked questions

1Which curve model should I choose?
Weibull (the default) if your business shows losses concentrated early, which is common in subscriptions; constant if your churn is genuinely stable month to month; two-phase if you know the month your churn changes regime — for example, when onboarding ends.
2Why isn't first-month retention 100%?
Month 1 already deducts the first cancellations: the projection applies the curve from the first period. The 100% corresponds to month zero, the moment of sign-up.
3What exactly is the half-life?
The month by which half the original cohort has canceled, linearly interpolated between the two months straddling the 50% crossing. It is to retention what the median is to the mean: robust and easy to compare.
4Does the LTV include revenue beyond the horizon?
No: it sums only the projected months. If you extend the horizon from 24 to 36 months, the bottom-up LTV grows with the curve's tail — that sensitivity to the horizon is information, not a defect.
5How do I compare this against my real data?
Build your real triangle (retention by age for each monthly cohort) and overlay it on the projected one: wherever your cohorts fall below the curve, there is a localized problem at that customer age.

Tools from the same topical cluster. Use them together to close the loop on your analysis.

Last updated: July 19, 2026

View methodology