Churn calculation is mathematically simple but has just enough edge cases to trip people up. The formula is straightforward; the discipline is in what you include and what you exclude.
The basic formula
Customer churn rate = customers lost during period ÷ customers at start of period
Revenue churn rate = MRR lost during period ÷ MRR at start of period
Both are typically expressed as a monthly percentage. For example, if you started the month with 1,000 subscribers and 50 canceled, monthly customer churn is 5%.
The rules that matter
- Exclude new acquisitions during the period. Otherwise the rate trends down whenever marketing has a good month, hiding real retention problems.
- Use start-of-period base, not average. Some methodologies use average customers during the period as the denominator. That is fine for a smoothing effect but harder to reconcile cohort-to-cohort.
- Count cancellations at the right moment. Subscribers who cancel mid-cycle but have access until the cycle ends — count them in the period they actually stop being customers, not when they clicked cancel.
- Separate voluntary from involuntary. Customer churn is one number; the split into voluntary (cancellations) and involuntary (payment failures) is the actionable detail.
Customer vs. revenue churn
If every customer pays the same amount, customer and revenue churn are the same number. When they diverge:
- Revenue churn higher than customer churn = you are losing your high-value subscribers faster. Bad news.
- Revenue churn lower than customer churn = your losses are concentrated in low-value plans. Less alarming.
Annualizing monthly churn
Common mistake: people multiply monthly churn by 12. That overstates the loss because the base shrinks each month. The correct conversion is 1 − (1 − monthly_churn)^12. A 5% monthly churn equals about 46% annual churn, not 60%. For the bigger picture see churn rate analysis.