What churn rate measures, and the two things it can count
Churn rate is the proportion of what you had at the start of a period that you no longer have at the end. The formula is trivial. Getting the two numbers into it correctly is where every mistake lives.
The first decision is what you count. Logo churn counts customers: 42 cancellations out of 1,200 accounts is 3.5%. Revenue churn counts money: $3,300 of lost MRR out of $96,000 is 3.44%. These two rates are almost never equal, and the gap between them is diagnostic. If revenue churn is much lower than logo churn, you are losing small accounts and keeping big ones — usually a sign that your product fits the upper half of your market better than the lower half. If revenue churn is higher, your largest customers are leaving, and that is an emergency dressed up as a small percentage.
The second decision is the denominator. The standard is customers at the start of the period, because churn is a property of a cohort that was already there. Using the average of the opening and closing balance is defensible in a slow-growing business and dishonest in a fast-growing one: adding 200 customers in the month you lost 100 pushes the denominator up and the reported churn down, with no improvement in retention whatsoever.
The third is the period. Churn is not a rate until you attach a window to it, and a 6% figure means completely different things monthly and annually. That is why this calculator normalises whatever you measured to a monthly equivalent before it does anything else.
Why you cannot just multiply monthly churn by twelve
Churn compounds against a shrinking base, so at any non-zero rate annual churn is less than twelve times monthly churn. If you lose 5% of customers every month, after twelve months you have not lost 60% — you have 0.9512 = 54.0% of the cohort left, so you lost 46.0%.
The correct conversion runs through the survival rate, not the loss rate. Retention for the year is monthly retention raised to the twelfth power, and annual churn is one minus that. Going the other way, from an annual figure to a monthly one, you take the twelfth root: m = 1 − (1 − a)1/12. A quarterly figure uses the cube root, a six-month figure the sixth root. Multiplying and dividing by twelve is only a passable approximation below about 1% monthly, and it becomes badly wrong above 5%.
The same compounding gives you the average lifespan. If a constant fraction m of your customers leaves each month, the expected number of months a customer stays is 1 ÷ m. At 5% monthly churn that is 20 months; at 2% it is 50 months; at 0.5% it is 200 months, or nearly 17 years. This single reciprocal is the bridge from churn to customer lifetime value, which is why a small change in churn moves LTV so violently — dropping monthly churn from 4% to 3% lifts average lifespan from 25 months to 33, a third more revenue per customer for no change in price.
Worked example: a month with 1,200 customers and $96,000 of MRR
A subscription business opens the month with 1,200 customers and $96,000 in MRR. During the month it loses 42 customers, wins 95, sees $3,300 of MRR cancel or downgrade, and books $2,100 of expansion from existing accounts.
- Logo churn. 42 ÷ 1,200 = 0.035 = 3.50% for the month. Retention is 96.50%.
- Annualise it. Monthly retention is 0.965. Raise to the twelfth power: 0.96512 = 0.6521. Annual churn is 1 − 0.6521 = 34.79%.
- Average lifespan. 1 ÷ 0.035 = 28.6 months, a little under two and a half years.
- Gross revenue churn. $3,300 ÷ $96,000 = 3.44%.
- Net revenue churn. ($3,300 − $2,100) ÷ $96,000 = $1,200 ÷ $96,000 = 1.25%.
- Net revenue retention. ($96,000 − $3,300 + $2,100) ÷ $96,000 = $94,800 ÷ $96,000 = 98.75%.
Now read the three revenue figures together. Gross revenue churn of 3.44% is slightly below logo churn of 3.50%, so the accounts leaving are marginally smaller than average — nothing alarming. Expansion recovers about two-thirds of the loss, leaving net revenue churn at 1.25% and NRR just under 100%. This business shrinks slightly on its existing base and must win new customers to grow at all. Push expansion up by another $1,200 a month and NRR reaches exactly 100%; anything beyond that and the installed base grows by itself.
How to read the result: what churn is acceptable
The only honest answer depends on who you sell to and what you charge, because churn scales with how easy the purchase was to make.
For B2B software sold to mid-market and enterprise buyers, practitioners generally treat annual logo churn in the mid single digits as strong and low double digits as tolerable. Convert before you compare: 5% a year is a monthly rate of 1 − 0.951/12 = 0.43%, while 12% a year is 1.06% a month. Annual contracts and procurement friction do much of the work here.
For self-serve monthly SaaS sold to small businesses, monthly churn of 3% to 5% is the common range, and below 2% is genuinely good. That is not slack management; a business with no contract and a credit-card checkout is structurally easier to leave.
For consumer subscriptions — media, apps, boxes — monthly churn frequently runs 5% to 10% and the whole model is built around it, with acquisition cost kept low enough to repay inside a handful of months.
Treat all three bands as rules of thumb rather than measured statistics — they are the ranges practitioners work with, not figures from a published survey. Your own trailing trend, converted to a monthly equivalent so the periods match, is worth more than any of them.
Whatever your segment, two derived numbers matter more than the level. First, is your average lifespan longer than your CAC payback period? If customers leave before they have repaid acquisition cost, growth destroys cash no matter how good the headline looks. Second, is net revenue retention above 100%? Negative net revenue churn means expansion outruns losses and the installed base compounds without a single new logo — the single most valuable property a subscription business can have.
Beware one artefact: churn measured over a short window on a small base is extremely noisy. With 80 customers, one cancellation is 1.25%. Report a three-month trailing rate until your base is a few hundred.
Monthly churn converted to annual churn and average lifespan
| Monthly churn | Monthly retention | Annual churn | Average lifespan |
|---|---|---|---|
| 0.5% | 99.5% | 5.84% | 200 months (16.7 yr) |
| 1.0% | 99.0% | 11.36% | 100 months (8.3 yr) |
| 2.0% | 98.0% | 21.53% | 50 months (4.2 yr) |
| 3.0% | 97.0% | 30.62% | 33.3 months |
| 5.0% | 95.0% | 45.96% | 20 months |
| 7.0% | 93.0% | 58.15% | 14.3 months |
| 10.0% | 90.0% | 71.76% | 10 months |
Average lifespan assumes a constant monthly hazard rate. Real cohorts churn hardest in months one to three, so this reciprocal flatters an early-life churn problem and understates the loyalty of survivors.
Customers won this period do not belong in either half of the fraction
A customer who signs up on the 8th and cancels on the 25th is a real loss, but including them makes the churn rate incoherent: they were never part of the cohort the denominator describes, and they had far less than a full period of exposure.
The clean convention is to measure churn only within the opening cohort, and to track early cancellations separately as a trial or onboarding failure rate. If you must include mid-period joiners, move to true cohort analysis — group customers by join month and track each group's survival — rather than patching the aggregate formula. Cohort tables also reveal something the aggregate rate hides completely: whether churn is concentrated in the first 90 days, which is a product-onboarding problem, or spread evenly, which is a value-delivery problem.
Mistakes that corrupt a churn number
- Including new customers in the denominator. Growth then masquerades as retention. Use the opening balance.
- Multiplying monthly churn by twelve. Overstates annual churn badly above 2% monthly, because losses compound against a shrinking base.
- Mixing voluntary and involuntary churn. A failed credit card is a payments problem with a payments fix; a cancellation is a product problem. Report them separately.
- Counting downgrades as churn in the logo rate. A customer who drops from 50 seats to 5 has not churned. Capture that in contraction MRR instead.
- Netting expansion into gross churn. Gross revenue churn must show the full loss. Net churn is a separate line, and hiding the gross figure hides the leak.
- Reporting monthly churn on a base of a few dozen. One cancellation swings the rate by whole percentage points. Use a trailing three-month figure.
- Timing annual contracts as monthly churn. If most customers can only leave at renewal, monthly churn is lumpy and the annual cohort rate is the meaningful measure.
Churn in the wider set of retention metrics
Churn is one of a family of four measures that describe the same reality from different angles, and mature reporting shows all four.
Customer retention rate is simply the complement of logo churn, and teams tend to prefer whichever framing makes progress feel more visible — moving retention from 95% to 96% sounds smaller than cutting churn from 5% to 4%, though they are the same event. Gross revenue churn puts a dollar weight on each departure. Net revenue churn subtracts expansion and answers whether the installed base is growing or shrinking. Net revenue retention restates that as a level rather than a loss, and is the figure investors quote most.
Downstream, churn feeds two calculations directly. Average lifespan drives lifetime value, and therefore every judgement about how much you can afford to spend on customer acquisition. Churn also sets the treadmill speed in any forward plan: your MRR only grows if new plus expansion revenue beats churned plus contracted revenue, which is exactly the arithmetic in a subscription revenue forecast.
One limit worth stating plainly. A single aggregate churn rate assumes every customer faces the same constant risk of leaving each month. They do not. Survival analysis on cohorts — or at minimum a churn-by-tenure table — is the right tool once you have a few thousand customers, because it separates an onboarding failure from a long-run value failure. The aggregate rate tells you the size of the leak; only cohorts tell you where it is.
Key terms
- Logo churn
- Churn counted in customers or accounts, regardless of how much each one pays. Also called customer churn or unit churn.
- Gross revenue churn
- MRR lost to cancellations and downgrades as a share of opening MRR. Never netted against expansion.
- Net revenue churn
- Gross revenue churn minus expansion MRR. A negative value means the installed base grew by itself.
- Contraction
- Revenue lost from a customer who stayed but bought less — fewer seats, a cheaper plan, a reduced usage tier.
- Involuntary churn
- Cancellation caused by a payment failure rather than a decision to leave. Report it separately from voluntary churn, because a card updater and a retry-and-email sequence recover a meaningful share of it while product change does not.
- Cohort
- A group of customers who joined in the same period, tracked over time. The only way to see whether churn is an early-life or a whole-life problem.