Renewal risk is a 90-day-warning problem disguised as a 30-day-surprise problem. The signals that predict churn are visible months before the renewal call, but they hide in places CSMs don’t monitor without a structured scan. Three patterns repeat across every CS book.
The original champion is the highest-value signal. A customer who renews you because the original buyer is still in the seat is a stable renewal. A customer where the original buyer left and was replaced by someone who didn’t sign the contract is a renewal-at-risk by structural definition — the new person inherits the budget line but not the conviction behind it. The prompt surfaces champion status first because nothing else matters as much.
Layoffs and lawsuits change the renewal conversation from “renew at growth” to “renew flat or down.” A customer in active cost discipline will look at every vendor line item. A customer in litigation will scrutinize every contract clause. The signals do not predict churn directly — they predict scope reduction, term reduction, and price pressure. All of which are renewal-risk events even when the customer technically renews.
Mixed-state accounts are the most important to surface. A customer with expansion signals AND risk signals firing is the account where the CSM’s executive sponsor needs to make a call. Lead with expansion and the risk gets ignored. Lead with risk and the expansion opportunity gets lost. The right move is to surface both at the same time so the strategy gets set deliberately. The cloud data platform row in the demo is the canonical case — three AI acquisitions building toward expansion AND a new CRO, layoffs, plus active securities litigation. Both are real. Both matter. Both inform the same conversation.
Stable accounts confirm the scan worked. A monthly run that flags 3 of 5 accounts is a working scan on a noisy book. A monthly run that flags 5 of 5 every month means the trigger threshold is too loose. A monthly run that flags 0 of 5 every month means the scan isn’t catching real signals. The prompt’s output distribution is itself a calibration signal for the CS leader.
Data drawn from Lusha’s signals layer, built on 300M+ verified contacts and millions of company records under GDPR, CCPA, SOC 2, ISO 27701, ISO 31700, and TRUSTe.