Churn rate

A 5% monthly churn rate and a 5% monthly churn rate can be two entirely different businesses. In one, most of the loss is cards that expired. In the other, most of it is people who watched what they came for and left in week three.

Three splits do most of the diagnostic work. The first separates voluntary from involuntary churn, because a failed payment answers to a recovery sequence and a deliberate cancel answers to the product or the price. The second is by tenure, which usually reveals that the number is dominated by a single early window rather than spread evenly across the base. The third distinguishes subscriber churn from revenue churn: losing an annual subscriber and losing a monthly one count the same in the first and are nowhere near equal in the second, which is why revenue churn can climb in a month when the subscriber count looks stable.

Period and denominator have to be stated too, or the metric stops being comparable to its own history. Monthly churn on a base that is mostly annual plans is measuring a small slice of the base each month, since only the cohort reaching renewal can churn at all. Cohort views handle that honestly by following one group of subscribers through their own timeline instead of averaging across everyone. Reading churn against what those subscribers were actually shown, and when, is where the number stops being a scoreboard and starts pointing at something changeable, which is the pairing Insights is built for. Alongside ARPPU, it is also the other half of any lifetime value estimate.