North Metric scores every customer daily on the signals Stripe already holds, adds your product usage when you connect it, and hands you the accounts to call this week with the reason and the dollars attached.
Works from Stripe alone on day one · Add PostHog, Segment or our SDK for earlier warning · No CS seats
Most customers who leave never complain. The signals were there for weeks: a downgrade, a card that bounced, a login that stopped. Nobody was watching every account.
Scheduled cancellations, past-due invoices, downgrades, interval switches, expiring cards — scored daily, ranked by MRR.
Stripe has the events. It doesn't compute churn by plan, by cohort, or by whether the customer chose to leave. So the answer lives in a spreadsheet you rebuild every board meeting.
Churn by plan, by cohort, by voluntary versus involuntary, across 15 dimensions. Updated daily.
Machine-learning churn models need thousands of churn events and a product-analytics pipeline. You have a few hundred customers and a Tuesday morning.
A weighted-signal score, not a black box. Works from a few hundred customers. Add usage data later for earlier warning.
Every night, Churn Radar scores each account on billing signals that lead churn by weeks: a scheduled cancellation, a past-due invoice, a downgrade, a switch from annual to monthly, discount dependence, tenure under 90 days, an expiring card.
Connect product usage and the score also sees logins, active seats and feature depth. The result is a ranked list, not a black box: each account shows the signals that fired and the MRR at stake.
Most health scores need four data sources and a customer-success team to maintain them. North Metric’s starts from billing: tenure, payment health, plan movement, discount depth. That is two of the four classic signal groups, with no tracking code.
Add product usage and the weights learn from your own churned accounts which signal dropped first. The company-level score, 0 to 100 with a letter grade, tells you which dimension to fix first.
Cohort Decay watches each signup month’s retention curve and flags the ones falling faster than your average, with the plan and channel patterns behind them.
It is how a pricing change or an onboarding regression shows up before it shows up in the churn rate.
Whale Alert measures revenue concentration and names the accounts that would move your MRR if they left. It watches those accounts harder, and it tells you when concentration crosses the line where one cancellation becomes a bad quarter.
When someone does leave, Win-Back Engine ranks churned customers by former MRR and likelihood of return, separates the ones who left on a failed card from the ones who chose to go, and drafts the note.
Our churn formula excludes same-period joins and reactivations, so it can read lower than other tools’, and for a clearer reason.
Weighted signals capped at 1.0. A scheduled cancellation and a past-due invoice are the heaviest. Downgrades, interval changes, discount dependence, tenure under 90 days, and an expiring card add. Usage signals join when connected.
Your rate against the P50 for your MRR tier. Above P50 is flagged. A spike is two standard deviations above your own last twelve months, needing six months of history.
Churn across 15 dimensions. A segment is flagged above twice your overall P50, critical above three times.
0 to 100, weighted: benchmark gap 25, net retention 20, gross retention 20, concentration 15, churn trend 20. Missing components redistribute. Grades: A from 85, B 70, C 55, D 40. Withheld with too little data.
P25, P50, P75 for your MRR tier, drawn from the North Metric benchmark database. Your position relative to peers, not to an absolute threshold.
It scores and estimates, it does not predict. Daily, not real time. No product usage until connected. Never writes to Stripe.
Churn management is the work of finding which customers are at risk of leaving, understanding why, and acting before they go. It covers voluntary churn, where the customer decides to cancel, and involuntary churn, where a failed payment ends the subscription without a decision.
| Feature | CS platforms | Cancel-flow tools | Analytics | North Metric |
|---|---|---|---|---|
| Built for | CS teams with seats | High-volume consumer | Reporting | Founders, no CS team |
| Data | Usage, CRM, support, billing | Billing | Billing | Billing + usage when connected |
| Output | Health colours and playbooks | Save offers at cancel | Charts | Named accounts, reasons, dollars |
| Time to signal | A quarter | At cancellation | An hour | Under two minutes |
| Price | $12K/yr and up | $200–$825/mo | $99–$1,000/mo | $79/mo flat |
During your 14-day Pro trial, three agents — Churn Radar, Payment Recovery, and Upgrade Finder — read your Stripe data daily and price every risk and opportunity in dollars. If they don't surface at least 12.6% of your current MRR in actionable opportunities, we extend your trial automatically. If we still haven't found it, your first month of Pro is on us.
Bring your Stripe account. Twenty minutes, and you leave with the accounts to call this week.
Read-only access · No credit card · Agents trial 14 days