Customer churn rate, also called logo churn, is the percentage of paying customers a business loses over a period. The standard formula is customers lost during the period ÷ customers at the start of the period × 100.
That formula is one defensible choice, not the only one. Divide by the average customer count instead, or follow a single signup cohort, and the same month produces a different number. Annualise a monthly rate by multiplying by twelve and the result is wrong again. Churn rate is only comparable once the denominator, the period and the edge cases are fixed and written down.
The churn rate formula, and its three denominators
The numerator is the easy part: paying customers who cancelled or did not renew. The disagreement is about what to divide by. Three versions are in common use, and each answers a slightly different question.
| Method | Denominator | Question it answers | Watch for |
|---|---|---|---|
| Start of period | Paying customers on day one of the period | What share of the customers I started with did I lose? | Customers who join and leave inside the period are outside the starting list, so they do not count |
| Average customers | The average of the customer count at the start and end of the period | What share of my typical customer base left? | Fast growth raises the denominator and lowers the rate even if the same customers leave |
| Cohort | Customers who started in one specific month or quarter, tracked over time | How quickly does a given group of customers decay? | Produces one curve per cohort, not one company-wide number |
Paddle describes the average method this way: "You divide your number churned by the average of your customer count between days 1 and n." Some billing tools use a fourth variant that adds new customers to the starting count. It is the most forgiving of the four, because every customer won during the period enlarges the denominator.
None of these is wrong. The start-of-period method is the easiest to audit, because the customer list is frozen on day one and every number traces back to a named account. The mistake is mixing them, or comparing a figure calculated one way with a figure calculated another.
A worked example: one month, three answers
A company starts the month with 400 paying customers. During the month it wins 40 new customers. 12 customers cancel: 10 from the starting list and 2 who signed up that month and left before it ended. It finishes with 428.
| Method | Calculation | Churn rate |
|---|---|---|
| Start of period | 10 ÷ 400 | 2.5% |
| Starting plus new customers | 12 ÷ 440 | 2.7% |
| Average customers | 12 ÷ 414, where 414 = (400 + 428) ÷ 2 | 2.9% |
The spread from 2.5% to 2.9% comes entirely from method. Nothing about the customers changed. In a board deck, a benchmark comparison or a quarter-over-quarter trend line, that gap is large enough to be mistaken for a real improvement or a real problem.
Monthly to annual churn: why not multiply by twelve
Churn compounds. Each month's rate applies to the customers still left, not to the original base, so a constant monthly rate removes fewer customers in absolute terms every month.
Annual churn = 1 − (1 − monthly churn)12
At 3% a month, a cohort keeps 0.9712 of its customers after a year, so annual churn is about 30.6%. Multiplying by twelve gives 36%, which overstates the annual loss by more than five points. The error grows as the monthly rate rises.
The conversion also works in reverse, and the same rule applies: never divide an annual rate by twelve to get a monthly one. When a figure is quoted without saying whether it is monthly or annual, it cannot be compared with anything until that is settled.
Cohort churn and what a blended rate hides
A single company-wide monthly churn rate blends customers who signed up last month with customers who have been paying for years. Those groups rarely leave at the same speed. Cohort churn separates them by tracking each signup group on its own.
| Month since signup | Customers remaining | Lost that month | Cohort monthly churn |
|---|---|---|---|
| 0 | 40 | ||
| 1 | 36 | 4 | 10.0% |
| 2 | 33 | 3 | 8.3% |
| 3 | 32 | 1 | 3.0% |
This cohort lost 20% of its customers in three months and kept 80%. The company's blended monthly figure over the same stretch could still read 2.5%, because the older base outnumbers the new cohort and leaves slowly.
Two consequences follow. A blended rate can rise when acquisition speeds up, simply because more customers are in their riskiest months, even if no cohort got worse. And a blended rate can hold steady while every new cohort decays faster than the last. Read the blended figure for the headline, and the cohort curves for the cause.
Lock the definition before you compare
Most arguments about churn are definition arguments. Decide each of these once, write the answers next to the metric, and apply them every period.
Customers who join and leave in the same period
Under the start-of-period method they never enter the calculation. Under the other methods they count. Pick one treatment and keep it, because a month with heavy acquisition will otherwise swing the rate on its own.
Reactivations
A customer who cancels and returns within the same period can be netted out or counted as churned and then won again. Netting makes churn look lower. Whichever you choose, apply it the same way in every period.
Partial cancellations
A customer who drops one of two subscriptions but keeps paying is still a customer. That is a revenue loss, not logo churn, and it belongs in gross revenue retention.
Free trials and involuntary churn
Count only paying customers, so a trial that never converts is not churn. Failed payments that are never recovered are churn, but report them separately as involuntary churn, because the fix is billing, not the product.
The written definition should state the method, the period (monthly or annual), and the treatment of each case above. Once it is fixed, the number can be compared with last quarter. Until then, it cannot.
Customer churn vs revenue churn
Customer churn counts accounts. Revenue churn counts money, so one large cancellation can move revenue churn sharply while barely changing logo churn, and the reverse is true for many small ones. Revenue-based retention, with and without expansion, is covered in gross revenue retention and net revenue retention.
Frequently asked questions
How do you calculate customer churn rate?
Divide the number of paying customers lost during the period by the number of paying customers at the start of the period, then multiply by 100. Some teams divide by the average customer count instead. Either is defensible, but state which one you use, because they give different answers for the same month.
How do I convert monthly churn to annual churn?
Use annual churn = 1 − (1 − monthly churn)12. A 3% monthly churn rate is about 30.6% a year, not 36%. Multiplying by twelve ignores that each month's churn applies only to the customers who are still left.
Should new customers be included in the churn rate denominator?
In the standard start-of-period method, no: the denominator is the customers you had on day one. Average-based methods include new customers indirectly through the end-of-period count. Adding all new customers to the denominator gives the lowest rate of the common methods, so it should be labelled clearly if used.
What is the difference between logo churn and customer churn?
None. Logo churn, customer churn and subscriber churn all count the number of paying customers lost. They differ from revenue churn, which measures the recurring revenue lost and can move very differently when large and small customers leave at different rates. Revenue-based retention is covered in gross revenue retention.
What is cohort churn?
Cohort churn tracks one group of customers who started in the same month or quarter and measures how many leave in each period after signup. It shows how fast a given group decays, which a single blended monthly rate cannot, because the blended rate mixes new and long-standing customers.
Is churn rate the same as retention rate?
For customers, they are two sides of the same calculation when measured on the same starting list: customer retention rate is 100% minus customer churn rate. Revenue retention is different, because expansion and downgrades change the result without any customer leaving.