Retention and Churn: Are Customers Staying?

What are retention and churn?

Retention measures how much of a customer group or its revenue remains over time. Churn measures loss. Together they help explain whether customers continue receiving value and whether growth lasts.

Define the population, period, and event before calculating a rate. Losing a free trial user differs from losing a paying account. A cancelled subscription may remain active until its paid term ends.

Measure	What it tracks
Customer retention	Starting customers still retained at the end
Customer churn	Starting customers lost during the period
Gross revenue retention (GRR)	Starting recurring revenue left after churn and contraction, excluding expansion
Net revenue retention (NRR)	Starting recurring revenue left after churn, contraction, and expansion

For revenue retention, exclude revenue from new customers. Stripe explains the difference between GRR and NRR.

Why they matter

New sales can hide a weak customer base. A business may grow total revenue while repeatedly losing customers it paid to acquire. Retention helps you investigate onboarding, customer fit, product value, service quality, and payment failures.

Customer and revenue retention answer different questions. Losing a few large accounts can have a much greater revenue effect than losing many small ones. Expansion can also hide meaningful customer loss.

Worked example: one month of subscription changes

All figures are invented for teaching, not benchmarks or forecasts. Begin the month with 100 paying customers and $10,000 monthly recurring revenue (MRR). During the month:

• Ten starting customers leave, removing $1,500 MRR.
• Other starting customers downgrade, removing $500 MRR.
• Retained starting customers expand by $2,500 MRR.
• Twenty new customers add $2,000 MRR.

Assume no reactivations or other changes.

Calculation	Result
Customer churn: 10 ÷ 100	10%
Customer retention: 90 ÷ 100	90%
GRR: ($10,000 − $1,500 − $500) ÷ $10,000	80%
NRR: ($10,000 − $1,500 − $500 + $2,500) ÷ $10,000	105%
Ending MRR including new customers	$12,500
Ending customer count: 100 − 10 + 20	110

Total MRR grew 25%, yet ten starting customers left. NRR above 100% shows expansion within the starting base exceeded its revenue losses. It does not establish that the lost customers were satisfied or that retention is healthy in every segment.

GRR excludes expansion, so it cannot exceed 100% under this definition. Stripe discusses how customer loss can affect revenue differently.

Use cohorts to see the pattern

A cohort groups customers sharing a start period or another relevant characteristic. Follow the same group over time instead of mixing new signups into the retention denominator.

Acquisition cohort	Started	Active at month one	Active at month three
January	50	40 (80%)	30 (60%)
February	40	34 (85%)	28 (70%)

These invented results suggest a question about improvement, not a causal conclusion. Compare customer mix, channel, product version, and onboarding. Do not report month-three retention for a cohort that has not yet had three months to mature.

For an infrequently used product, choose a window that matches the real customer need. Subscription renewal, active usage, and successful outcomes are separate measures; report each clearly.

Build a consistent measurement policy

1. Define the account or customer identifier and remove duplicates.
2. Set opening and closing dates and a consistent currency policy.
3. Define cancellation, inactivity, nonpayment, downgrade, and reactivation.
4. Separate customer counts from recurring revenue movements.
5. Reconcile the revenue bridge to billing records.
6. Review segments and cohorts alongside the total.

Monthly rates do not become annual rates by simple multiplication. Under an illustrative constant 5% monthly churn assumption, annual retention is 0.95 to the power of 12, about 54%. Real cohorts may behave differently as they age or renew.

Investigate the cause before choosing a fix

Pattern	Useful investigation
Customers never activate	Observe onboarding and check whether the offer attracted the right buyer
Cancellation after a few uses	Ask whether the outcome recurs and whether value lasts
Usage continues but payment fails	Review billing and recovery processes
Downgrades in one segment	Check package fit, budget changes, and product value
Strong NRR but falling customer count	Check concentration and dependence on expansion

Interview lost customers without steering them toward your preferred explanation. Record the reason, evidence, owner, and next test.

What investors are looking for

Be ready to explain the definitions, customer history, renewal schedule, and concentration behind the headline. Show how recent cohorts compare with earlier ones and how retention affects CAC payback and cash needs.

Benchmarks depend on customer size, contract length, model, and measurement period. A short history should be labeled clearly. Avoid presenting a few early renewals as evidence of years of retention.

Common mistakes

• Counting new customers as retained customers.
• Including new-customer MRR in NRR.
• Mixing usage retention and subscription retention.
• Hiding customer losses behind expansion.
• Comparing monthly and annual rates directly.
• Filling immature cohort periods with assumed results.

Founder checklist

Document your population, period, event definitions, and revenue basis. Calculate customer retention, GRR, and NRR where relevant. Review cohorts and concentration. Identify the largest loss mechanism and assign a focused experiment.

Your next action: reconcile one period's opening revenue, losses, expansion, and new sales before interpreting the growth rate.

Sources and further reading

• Stripe: Net Revenue Retention
• Stripe: Customer Churn

References checked November 10, 2026. Worked examples are simplified teaching scenarios.