How to calculate churn rate: the short version
To calculate churn rate, divide what you lost during a period by what you started the period with. The simplest customer churn formula is:
Customer churn rate = customers lost during the period ÷ customers at the start of the period
If you began March with 400 paying customers and 18 of them canceled during March, your monthly customer churn rate is 18 ÷ 400 = 4.5%.
That is the right starting point, but "churn rate" is used for at least three different measurements: customer (logo) churn, gross revenue churn, and net revenue churn. They can move in opposite directions, and tools calculate them with different denominators. This guide shows each formula, works through one dataset end to end, and lists the decisions you need to document so your churn rate is comparable from month to month.
You can check every formula below with the SaaS churn and retention calculator.
The three churn rates you should calculate
1. Customer churn rate (logo churn)
Customer churn counts accounts, not money:
Customer churn rate = churned customers ÷ customers at start of period
ChartMogul's SaaS metrics cheat sheet uses this start-of-period definition. Recurly's churn rate benchmarks page describes the same calculation as subscribers lost divided by subscribers at the start of the period.
Customer churn tells you how many relationships you are losing. It treats a $9 customer and a $900 customer the same, which is useful for product and support questions but misleading for financial ones.
2. Gross revenue churn rate (gross MRR churn)
Gross revenue churn measures recurring revenue lost from existing customers, including downgrades:
Gross MRR churn rate = (churned MRR + contraction MRR) ÷ MRR at start of period
ChartMogul defines gross MRR churn as churn plus contraction MRR divided by MRR at the start of the period. Stripe's Billing analytics documentation reports a similar "churned revenue" figure as churned MRR plus contraction MRR. Gross revenue churn can never be negative, because expansion is ignored. That makes it the cleanest measure of how much of the existing base leaks out.
3. Net revenue churn rate (net MRR churn)
Net revenue churn offsets losses with expansion and reactivation from existing customers:
Net MRR churn rate = (churned MRR + contraction MRR − expansion MRR − reactivation MRR) ÷ MRR at start of period
This follows ChartMogul's description of net MRR churn. Net churn can be negative: if existing customers expand by more than others cancel or downgrade, the existing base grows on its own. Negative net churn is the same idea as net revenue retention above 100%, which is covered in the net revenue retention guide.
Worked example: one month, three churn rates
A B2B tool starts April with 400 customers and $24,000 of MRR. During April:
- 20 customers cancel, representing $1,000 of MRR;
- 15 customers downgrade, reducing MRR by $300;
- 25 customers upgrade or add seats, adding $1,100 of MRR;
- 2 former customers return, adding $100 of MRR;
- 40 new customers sign up, adding $2,400 of MRR.
Customer churn rate
20 ÷ 400 = 5.0%
Gross MRR churn rate
($1,000 + $300) ÷ $24,000 = $1,300 ÷ $24,000 ≈ 5.4%
Net MRR churn rate
($1,000 + $300 − $1,100 − $100) ÷ $24,000 = $100 ÷ $24,000 ≈ 0.4%
Ending MRR
$24,000 + $2,400 + $1,100 + $100 − $300 − $1,000 = $26,300
New customers are excluded from every churn numerator and denominator in this method. They did not exist at the start of the period, so they cannot have churned from the starting base. If you mix them in, a strong acquisition month will make churn look better than it is.
Notice what the three rates say together: the business lost one in twenty customers, lost a bit more than 5% of its starting revenue to cancellations and downgrades, but almost entirely replaced that loss through upgrades. The customers who leave are slightly more valuable than average (5.4% of revenue versus 5.0% of customers), which is worth investigating.
Different denominators, different answers
The start-of-period formula is the most common, but not the only one. Stripe calculates subscriber churn rate as churned subscribers in the past 30 days divided by active subscribers 30 days ago plus new subscribers in the past 30 days. Its documented example: 1,000 subscribers at the start, 100 added, 100 churned gives 100 ÷ (1,000 + 100) = 9.1%. Under the start-of-period method the same data gives 100 ÷ 1,000 = 10.0%.
Other analysts divide by the average customer count during the period:
Churn rate = churned customers ÷ ((start customers + end customers) ÷ 2)
None of these is wrong. The rule is to pick one, write it down, and never compare a churn rate calculated one way with a benchmark or a prior period calculated another way. When a buyer asks for your churn, state the formula alongside the number.
Converting monthly churn to annual churn
Monthly churn compounds; you cannot multiply it by 12. Customers who survive month one are exposed to churn again in month two, so:
Annual retention = (1 − monthly churn rate)^12
Annual churn = 1 − (1 − monthly churn rate)^12
At 2% monthly churn: 1 − 0.98^12 = 1 − 0.785 ≈ 21.5%. Recurly's benchmarks page makes the same point: 2% monthly churn is roughly 22% annual churn, not 24%.
The reverse conversion is:
Monthly churn = 1 − (1 − annual churn)^(1/12)
Stripe's cohort example shows the compounding visually: at 10% monthly subscriber churn, a cohort retains 90% after one month, 81% after two, 73% after three, 66% after four, and 59% after five.
Churn rate and customer lifetime
A rough lifetime estimate follows from a constant churn rate:
Average customer lifetime (months) ≈ 1 ÷ monthly customer churn rate
LTV ≈ ARPA ÷ monthly customer churn rate
Stripe uses this approach for subscriber lifetime value, dividing ARPU by subscriber churn rate. In its example, $500 ARPU and 9% churn produce about $5,555. At 5% monthly churn the average lifetime is roughly 20 months; at 2% it is roughly 50 months. These are simplifications: real churn is usually higher in the first months and lower for older cohorts. Use the LTV:CAC calculator to see how sensitive lifetime value is to small churn changes.
Decisions to document before you calculate
Churn rates are only comparable when the inputs are defined the same way each time.
- When is a customer churned? On the cancellation request, at the end of the paid term, or when a subscription becomes unpaid? Stripe treats a subscriber as churned when all subscriptions are canceled or marked unpaid.
- Voluntary versus involuntary. Recurly separates voluntary churn (active cancellation) from involuntary churn (loss after failed payment). Track both; the fixes differ completely.
- Accounts with several subscriptions. Count churn at the customer level, not the subscription level, unless you say otherwise.
- Annual plans. Annual customers can only churn at renewal. A business with many annual plans will show low monthly churn most months and spikes at renewal dates; measure annual cohorts at renewal instead.
- Trials and free plans. Exclude them from the starting base. A trial that does not convert is a conversion problem, not churn.
- Reactivations. Decide whether a returning customer offsets churn (net) or is ignored (gross).
- Pauses. Treat paused accounts consistently, usually as contraction or churn.
Cohort churn beats a single average
A blended monthly churn rate hides the shape of retention. Group customers by the month they started and track what share of each cohort, and of its starting MRR, remains after 1, 3, 6, and 12 months. Cohorts reveal whether recent product or pricing changes improved retention, and whether early-life churn is the real problem.
For context on what churn rates look like across price points and segments, see what is a good churn rate for SaaS. For the movement categories that feed these formulas, see what is MRR.
Common churn calculation mistakes
Including new customers in the denominator without saying so. It lowers churn in high-growth months.
Reporting only net revenue churn. Expansion can hide a serious cancellation problem. Always show gross churn too.
Multiplying monthly churn by 12. Use the compounding formula.
Using payment data as a proxy for subscriptions. A missing payment in a given month can mean an annual customer, a failed card, or a cancellation. RevenueBug's public trailing 30-day revenue, for example, is calculated from payments minus refunds and disputes, so it reflects revenue movement but is not itself a churn calculation; see the methodology.
Ignoring involuntary churn. Failed payments are often recoverable with retries and card-update reminders.
Measuring too small a base. With 30 customers, one cancellation moves churn by more than three points. Use longer windows or cohorts.
Frequently asked questions
What is the basic churn rate formula?
Customers lost during the period divided by customers at the start of the period. For revenue churn, replace customer counts with MRR lost to cancellations and downgrades.
Should I calculate churn monthly or annually?
Monthly is common for monthly-billed products; annual or cohort-at-renewal measurement is more meaningful when most customers are on annual plans. Convert between them with the compounding formula, not by multiplying or dividing by 12.
What is the difference between gross and net revenue churn?
Gross revenue churn counts only losses (cancellations and downgrades). Net revenue churn subtracts expansion and reactivation from those losses, so it can be negative.
Do downgrades count as churn?
They count in revenue churn as contraction, but not in customer churn, because the customer is still active.
What does negative churn mean?
Negative net revenue churn means expansion from existing customers exceeded the revenue lost from cancellations and downgrades, so the existing base grew without any new customers.
How do I calculate churn with very few customers?
Use a longer period, such as a quarter, or track cohorts. Report the raw counts alongside the percentage so readers can see how small the base is.