Churn is the percentage of customers, or revenue, you lose in a period. It's the leak in the bucket: if it's high, no amount of new users will fill it. You can see why in The Math of $1M MRR.
Customer churn
customer churn % = customers lost in period ÷ customers at start of period × 100
Example: you start March with 200 paying customers. During March, 10 cancel and 30 new ones sign up.
churn = 10 ÷ 200 × 100 = 5%
The 30 new customers don't go in the formula. That's the most common mistake (more below).
Retention is the flip side: 100% − churn = 95%.
Check your own numbers with the Percentage Calculator.
Revenue churn
Customer churn treats every account as equal. Revenue churn weights them by what they pay.
gross revenue churn % = (MRR lost to cancellations + MRR lost to downgrades) ÷ MRR at start × 100
Example: start of month MRR is 10,000. Cancellations lose 400, downgrades lose 100.
gross revenue churn = 500 ÷ 10,000 × 100 = 5%
If one customer paying 2,000 leaves, you might lose 1 of 200 customers (0.5% customer churn) but 20% of revenue. Track both.
Net revenue retention (NRR)
Existing customers also upgrade. NRR captures that:
NRR % = (start MRR − churned MRR − downgrade MRR + expansion MRR) ÷ start MRR × 100
Same example, plus 800 in upgrades:
NRR = (10,000 − 400 − 100 + 800) ÷ 10,000 × 100 = 103%
Over 100% means your existing customers grow revenue even with zero new sales. That's the strongest position a subscription business can be in. It's much easier to reach in B2B (seats, usage tiers) than in consumer apps, where plans rarely expand.
Monthly vs annual churn
You can't multiply monthly churn by 12:
annual churn = 1 − (1 − monthly churn)^12
| Monthly churn | Annual churn (not ×12) |
|---|---|
| 1% | 11.4% |
| 2% | 21.5% |
| 3% | 30.6% |
| 5% | 46.0% |
| 8% | 63.2% |
| 10% | 71.8% |
Each month's churn applies to a smaller group, so the annual figure is less than 12×. It's still brutal: 5% monthly churn means losing nearly half your customers every year.
Customer lifetime from churn
A rough estimate of how long a customer stays:
average lifetime (months) ≈ 1 ÷ monthly churn
At 5% monthly churn, about 20 months. This feeds straight into lifetime value. See LTV:CAC Ratio Explained.
Four mistakes that hide churn
- Counting new customers in the base. Dividing by start + new customers makes churn look smaller during fast growth, exactly when you most need the truth.
- Mixing monthly and annual plans. An annual subscriber can only churn at renewal. Report them separately, or annual plans will make monthly churn look artificially low for 11 months.
- Counting trials as customers. Trial-to-paid conversion is its own metric. Churn should start from the first payment.
- Ignoring involuntary churn. Failed card payments can be a big share of cancellations. Retrying payments and sending "update your card" emails recover some of it, which may be the cheapest growth you'll find.
Cohorts show what averages hide
A single churn number blends old and new customers. A cohort table groups customers by the month they joined and tracks what share are still paying:
| Joined | Month 1 | Month 2 | Month 3 | Month 6 |
|---|---|---|---|---|
| Jan | 100% | 78% | 70% | 61% |
| Feb | 100% | 82% | 75% | — |
| Mar | 100% | 85% | — | — |
(Illustrative numbers.)
If later cohorts retain better, your product changes are working. If every cohort's curve flattens out at some level, you've found the core of users who really value the product. That's a sign of product-market fit.
What's a "good" churn rate?
It depends heavily on who you sell to and at what price. Enterprise software with annual contracts churns far less than a cheap consumer app paid monthly. Rather than chasing someone else's benchmark:
- compare your churn with your own past months;
- compare cohorts with each other;
- and check that churn is low enough for your unit economics to work.