How to Calculate Churn Rate: Formulas, Denominators and Worked Examples

Churn rate is the share of customers you lose over a period. The formula fits on one line, but the result depends on three choices most guides skip: which customers go in the denominator, how you turn a monthly rate into an annual one, and what counts as "lost" when subscribers can pause, skip or fail a payment. This guide gives the formulas, works each one through with numbers, and shows how Stripe, Chargebee and Recurly make those choices. By industry: subscription churn for supplement brands, subscription churn for skincare brands, subscription churn for online pharmacies, subscription churn for telehealth companies, subscription churn for fitness brands and subscription churn for info product businesses.

What is churn rate?

Churn rate measures the customers, subscriptions or recurring revenue that end during a period, as a share of what you had when the period began. There are three common versions, and they answer different questions:

The difference between the first two matters as soon as a customer can hold more than one subscription. Stripe Billing counts a subscriber as churned only when all of their subscriptions are canceled or marked unpaid.1Stripe Docs, Billing analytics: metric definitions (MRR, subscriber churn rate, churned revenue, retention by cohort) Recurly makes the same distinction: a subscriber with two subscriptions can cancel one and still not count in subscriber churn.6Recurly Docs, Subscriber churn analysis Chargebee reports subscription churn, which counts each canceled subscription.3Chargebee Docs, Churn metrics (Billing 2.0, RevenueStory)

The churn rate formula

The classic formula, used by Recurly and in Chargebee's subscription churn reports:6Recurly Docs, Subscriber churn analysis3Chargebee Docs, Churn metrics (Billing 2.0, RevenueStory)

Customer churn rate = customers lost during the period ÷ customers at the start of the period × 100

The customers lost are those who were active on day one and are gone by the last day. Recurly's definition says exactly that: a subscriber churns if they weren't expired at the beginning of the period but are expired by the end of it.6Recurly Docs, Subscriber churn analysis

The same rule applies to any period. A quarter that starts with 2,000 customers and loses 240 of them has a 12% quarterly churn. Compare like with like: a quarterly rate next to a monthly rate means nothing until you convert one of them (the conversion is two sections down).

Which denominator? Start, average or end of period

The formula above is not the only one in use. Billing platforms publish different definitions, and the same month gives a different churn rate in each of them. Here's how each one treats customers who join during the period:

Now apply all four to the same month. Take the October example: 2,000 subscribers at the start, 400 new, 90 starting subscribers lost and 20 new subscribers lost, so 110 losses in total and 2,290 subscribers at the end.

The last row is a mistake, not a method: it counts the 20 new customers in the numerator but not in the denominator, so it inflates churn. The spread between the others is small in a calm month and grows with acquisition. A brand that doubles its sign-ups during a promotion will see its end-of-period churn drop while nothing changed for existing subscribers.

What to do with customers who join and leave in the same period? In the start-of-period method, they don't count: they belong in a cohort report or a first-month conversion metric. Chargebee excludes them explicitly, and Stripe includes them.3Chargebee Docs, Churn metrics (Billing 2.0, RevenueStory)2Stripe Support, Calculating churn rate in Billing If your first month is where most subscribers drop, track that drop on its own; don't let it disappear or swamp the monthly rate.

Monthly vs annual churn: why you can't multiply by 12

Churn compounds. In month two you lose a share of the customers left after month one, not of the original base. Multiplying a monthly rate by 12 overstates annual churn, and above about 8.3% a month it gives more than 100%, which is impossible.

Annual churn = 1 − (1 − monthly churn)^12

Monthly churn = 1 − (1 − annual churn)^(1/12)

Read the other way, an annual churn of 30% is a monthly churn of 2.93%, not 2.5%. In a spreadsheet, with the monthly rate in cell B2: =1-(1-B2)^12.

The formula assumes the same monthly rate all year, which is rarely true: churn is often highest in the first months after sign-up. For a real annual figure, measure it directly: customers at the start of the year who are gone by the end, divided by customers at the start of the year.

Revenue churn: gross and net

Customer churn treats a 15 USD subscriber and a 90 USD subscriber the same way. Revenue churn weights each loss by what it was worth. It is measured on MRR, and it ignores new customers: their revenue is growth, not retention.

Churned MRR comes from canceled or unpaid subscriptions. Contraction MRR comes from downgrades, smaller boxes or longer delivery intervals. Expansion MRR comes from upgrades and add-ons by existing subscribers. Stripe's "churned revenue" metric is churned MRR plus contraction MRR, which is the numerator of gross revenue churn.1Stripe Docs, Billing analytics: metric definitions (MRR, subscriber churn rate, churned revenue, retention by cohort) Chargebee also reports a narrower "cancellation MRR rate" that leaves downgrades out.3Chargebee Docs, Churn metrics (Billing 2.0, RevenueStory)

Back to October, now in dollars. Illustrative example: MRR on October 1 is 50,000 USD. The 90 subscribers who left were paying 2,000 USD a month in total. Other subscribers downgraded by 500 USD and upgraded by 1,200 USD. The 400 new subscribers added 9,000 USD, which stays out of every churn formula.

If upgrades had reached 3,000 USD, net revenue churn would be (2,500 − 3,000) ÷ 50,000 = −1%: negative churn, where the existing base grows even after losses. Gross revenue churn can't go below zero, which is why it's the stricter test. When and why to report each one is covered in gross vs net retention; how to build the MRR base itself is in MRR calculation, and what net revenue retention means as a ratio is in the glossary entry on net revenue retention.

Inside Tagada

Build the funnel once, route every customer to the right offer

Checkout, upsells and subscriptions on one flow, with conditions per country, order or customer.

Start free

Voluntary vs involuntary churn

Every churned subscriber left in one of two ways. Recurly's definitions: voluntary churn is subscribers who choose to leave; involuntary churn is subscribers lost to payment issues, such as a failed transaction and an unrecovered invoice.6Recurly Docs, Subscriber churn analysis Use the same denominator for both, and they add up to your total churn rate.

In October, 62 of the 90 lost subscribers canceled and 28 ended after renewals that were never paid:

The timing of involuntary churn trips up many reports. A failed renewal is not churn yet. Stripe keeps past_due subscriptions in MRR and among active subscribers, and counts churn only when the subscription is canceled or marked unpaid, one of the outcomes you can set for when the retry window ends.1Stripe Docs, Billing analytics: metric definitions (MRR, subscriber churn rate, churned revenue, retention by cohort) So the churn lands in the month dunning ends, not the month the card was declined.

Two edge cases to decide once and keep:

• Stop payment orders. When a cardholder tells their bank to stop paying you, the renewal fails, but the customer chose to leave. Classing it as voluntary keeps your involuntary churn a measure of payment problems.
• Subscribers still in recovery at month end. They're active until the window closes. Your latest month will look better than it will once retries finish, so compare complete periods.

Why renewals fail in the first place, by decline code, is in subscription payment failures. The recovery process, with retries, emails and grace periods, is in dunning and failed payment recovery.

Cohort churn: following one group of subscribers

A monthly churn rate mixes subscribers in their first month with subscribers in their third year. A cohort keeps them apart: everyone who started in the same month, followed month by month. Stripe assigns subscribers to cohorts by the month they first started generating positive MRR, and shows both subscriber retention and revenue retention per cohort; revenue retention can go above 100% because of expansion.1Stripe Docs, Billing analytics: metric definitions (MRR, subscriber churn rate, churned revenue, retention by cohort)

For a cohort, two numbers matter at each month n:

• Retention at month n = active cohort members at month n ÷ cohort size at start
• Churn in month n = cohort members lost during month n ÷ cohort members active at the start of month n

Illustrative example: 400 subscribers who started in January.

The shape is what a blended rate hides: in this example the cohort loses more in its first month than in the next three combined. If a newer cohort loses more in month one than this one did, look at what changed for it: the offer, the acquisition channel, the checkout, or the first renewal. Stripe's own documentation shows the compounding at work: with a steady 10% monthly churn, a cohort keeps this much of its starting subscribers:1Stripe Docs, Billing analytics: metric definitions (MRR, subscriber churn rate, churned revenue, retention by cohort)

How to build and read a cohort table is covered in what is cohort analysis.

Churn in subscription ecommerce: pauses, skips and reactivations

Physical subscriptions give customers more ways out than "cancel": skip a delivery, pause for two months, swap to a cheaper product, come back after a cancellation. Each one needs a rule before you can calculate churn. Here's how billing tools handle them, and a rule that keeps the number honest.

Two rules make pauses manageable. First, cap them: a pause with no end date is a cancellation that hasn't been admitted. Decide on a maximum, and count the subscriber as churned if the pause runs past it. Second, if your churn rate excludes pauses, show paused subscribers next to it. A rising pause count with a flat churn rate is a warning, not good news.

How to calculate churn in a spreadsheet

One row per month is enough. The columns below give the customer churn rate, the Stripe-style rate and the annualized rate, with a check that the counts reconcile.

Check column G against the real count of active customers at month end. If they don't match, some status changes (pauses, reactivations, refunds that ended a subscription) aren't classified yet, and the churn rate is wrong by that difference. Split column D into voluntary and involuntary to get the two rates from the previous section.

Common churn calculation mistakes

A few more that don't show up as a pattern:

• Mixing customers and subscriptions. Subscriber churn and subscription churn are different numbers when customers can hold several subscriptions. Name the one you report.
• Netting reactivations against churn. A customer who left in March and returned in May is a churn in March and a reactivation in May. Subtracting returns from losses hides both.
• Comparing your rate with a published benchmark. Benchmarks come from one vendor's customer base, with that vendor's definitions, often software companies on annual contracts. Compare your own cohorts over time instead.

Churn also feeds other metrics. Stripe estimates subscriber lifetime value as average revenue per user divided by the subscriber churn rate1Stripe Docs, Billing analytics: metric definitions (MRR, subscriber churn rate, churned revenue, retention by cohort), so an error in churn moves your lifetime value the opposite way. More on that in what is CLTV.

Where Tagada fits

Whatever tool you use, the method is the same: choose a denominator and keep it, compound monthly churn to get annual churn, report customer and revenue churn side by side, split voluntary from involuntary, and decide in advance how pauses, skips and reactivations count.