Churn and retention metrics
Two families of numbers describe whether customers stay: churn measures who left, and retention measures who stayed. Every rate on this page is measured against a cohort that already existed at the start of the period, so none of them can exceed 100%.
Churn rate: voluntary, involuntary, and the window-start cohort rule
Churn rate is the share of subscribers lost in a period. Voluntary churn is people who chose to cancel; Involuntary churn is people lost to failed payments. Both are measured against subscriptions that were already active when the period began, the window-start cohort, so the rate can never exceed 100%.
The denominator is every subscription active at the start of the window, from any source; the numerator is cancellations inside the window drawn only from that same starting cohort. A cancellation is involuntary when its reason is a failed or disputed payment, or when the subscription passed through past due or unpaid before it was cancelled; every other cancellation, including a detected subscription that simply lapses, is voluntary.
90-day retention
Of the customers who signed up, 90-day retention is the share still subscribed 90 days later. It comes from the cohort engine, pooled over recent cohorts, so changing the dashboard's date range does not change it.
SSA pools cohorts by first-payment month, newest first, going back up to 24 months, until it has at least 3 cohorts and at least 20 pooled starting subscribers; the rate is pooled retained divided by pooled starting subscribers. If those thresholds are never reached, the figure is blank with a reason shown instead. It is computed per source and never blended across sources on the dashboard hero.
Cohort retention (M1 / M3 / M6 / M12) and the heatmap
The Cohorts page groups customers by the month of their first successful payment and shows how many are still subscribed 1, 3, 6, and 12 months later. Darker cells on the cohort retention heatmap mean better retention; empty cells mean that month has not arrived yet.
A customer is retained at month N if they have any subscription that started on or before the first day of that month and either never cancelled or cancelled after it; a paused subscription still counts as retained. The starting count for each cohort is frozen the first time it is written, so a later data change cannot push retention above 100%.
Net revenue retention (live) / NRR
Net revenue retention (live) takes the revenue base you had at the end of last month and re-measures those same subscriptions today: a renewal counts in full, a cancellation counts as zero, and an upgrade or downgrade applies. Over 100% means your existing customers grew; brand-new customers are excluded on purpose.
The base is every subscription that counted toward MRR at the end of the previous month, valued at its amount then; the numerator is the same subscriptions' amount now, or zero if they no longer count toward MRR. It is computed per source and combined by summing the numerators and denominators.
Payment recovery rate and recovered revenue
Of the invoices that failed at least once and were retried, recovery rate is the share that eventually got paid, and recovered revenue is how much money those recoveries brought in.
The population is invoices with more than one payment attempt. The Payment Recovery report measures this over your selected range; a separately stored 90-day figure feeds the dashboard and digest. Only sources that report retry data contribute: a connector's invoices are modeled as single-attempt and never qualify, and Manual payments have no invoices at all.