Catching churn before
the cancel button.
A horizontal SaaS company at roughly $4M ARR, mid-market plan mix, lean customer success team. Anonymized at the customer's request.
Churn was being recognized after the fact.
The customer success team had clean usage signals — login cadence, feature engagement, support ticket trends — but no scalable way to act on them. By the time an at-risk account got a call, it was usually a save attempt, not a check-in.
Quarterly business reviews caught the same patterns, far too late.
A weekly check-in loop driven by product signals.
We built a weekly at-risk identification job that combined usage and support signals into a single scored list. A specialist agent drafted personalized check-in messages for each flagged account — referencing the actual feature drop or open ticket — and routed them through the CSM's Approvals Inbox before send.
Healthy accounts surfaced too: expansion candidates with rising adoption got a different lane drafted for the same CSM to review.
A measurable lift on accounts the team would have missed.
Within a quarter, the loop recovered 15–25% of flagged at-risk accounts that historically would have churned silently. Expansion conversations opened earlier on healthy accounts.
The CSM team reported spending less time on data pulls and more time on the conversations the data surfaced.