Time to churn is the number of months between a customer's start date and its cancellation. Reported as a median across churned accounts and paired with a survival curve, it answers a question that a churn rate cannot: when in the customer relationship the losses actually happen.
Median over mean
Take the churned accounts from a defined window, compute months of tenure for each, then take the median. A cohort where most cancellations land between months 9 and 15 with three accounts leaving after four years produces a median near 12 and a mean closer to 18. The median describes the typical exit. The mean describes a small group of long-tenured accounts that were never representative.
Report the count alongside the median. Twenty cancellations with a median of 11 months carries weight. Three cancellations with a median of 11 months is an anecdote, and the number will swing hard the next time a single account leaves.
Survival curves show the shape
A survival curve plots the share of a cohort still active at each month of tenure. Two companies can share an identical median time to churn and have completely different problems.
- Steep early drop, then flat: the base loses accounts that were mismatched at purchase or never reached working value, then holds. The fix sits in qualification and onboarding. - Steady decline across all tenures: the product keeps losing customers who did succeed at first. The fix sits in ongoing value delivery and competitive position.
The curve also prices retention work honestly. Flattening a first-year drop protects revenue for every future cohort, while a save play at month 30 protects one account.
Active customers are not zero-tenure customers
The common error is computing time to churn only on accounts that already cancelled. Every active customer is still accruing tenure with no cancellation date, and dropping them biases the estimate downward, sometimes severely. This is right-censoring, and survival analysis handles it by counting active accounts as observations that survived at least as long as their current tenure rather than throwing them out.
The practical version does not require statistical software. Build the survival curve by cohort, mark the last observed month for each cohort, and stop reading the curve past the point where the youngest cohort still has data. A company with 30 months of history cannot make a statement about month 48.
Time to churn feeds the retention side of planning directly. Pair it with net revenue retention to see both how much revenue survives and when the losses arrive, then carry the timing into the revenue forecast so churn lands in the periods it actually occurs instead of being spread evenly across the year.
Frequently Asked Questions
How do you calculate time to churn?
Subtract each churned account's start date from its cancellation date, then take the median across the group. Use the median rather than the mean, because a handful of long-tenured accounts that finally cancel will pull an average upward and make the base look more durable than it is.
Why does time to churn understate real customer lifespan?
Because it only counts customers who already left. Every active account is still accumulating tenure and has no cancellation date yet, a condition statisticians call right-censoring. Survival analysis handles it by including active accounts as censored observations instead of dropping them.
What does a survival curve add over a single number?
Shape. A curve that drops steeply in the first six months and then flattens is a fit and onboarding problem. A curve that declines steadily across all tenures is a value problem that keeps compounding. Both can produce the same median while pointing at completely different fixes.
How does time to churn relate to lifetime value?
Lifetime value multiplies revenue per period by expected lifespan, so the timing estimate sets the horizon. If real losses cluster at month 14 and the LTV model assumes a 40-month lifespan, the payback math is wrong and every acquisition decision built on it is too generous.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like time to churn into prescriptive action for your team.
Schedule a Demo