Churn announces itself 90 days early

No customer wakes up and cancels. By the time the “can we discuss our contract” email lands, the decision is weeks old. What looks like a sudden churn is almost always a slow one that nobody was watching.

Anatomy of a preventable churn

A real pattern, anonymized, from an account worth $48,000 ARR:

• Day 0: Weekly product usage drops 42%. A team lead left; nobody re-onboarded the replacements. Visible in product analytics. Nobody looks.
• Day 20: Three support tickets sit unresolved. Individually minor. Together, they read as “this vendor doesn’t care.” Visible in the help desk. Nobody connects them to the account.
• Day 45: Engagement drops 28%. The champion stops opening emails. The monthly check-in gets rescheduled twice. Visible in the CRM. Logged as “busy quarter.”
• Day 80: Procurement asks for a copy of the contract.
• Day 90: The renewal call becomes an exit interview.

Every signal existed. Each lived in a different system — product analytics, support desk, CRM — and no single person owned the composite picture. The churn wasn’t unpredictable. It was unread.

Health scores, not vibes

The fix is mechanical, not heroic:

1. Unify the signals. Usage, ticket state, engagement, sentiment — one score per account, weighted by what actually predicts churn in your base. Usage decline usually carries the most weight; unresolved tickets punch far above their size.
2. Trend it, don’t snapshot it. A health score of 48 matters less than the fact it was 82 ninety days ago. Direction and velocity are the alarm, not the level.
3. Alert inside the intervention window. A risk flag 90 days before renewal is a project. The same flag two weeks before renewal is a discount negotiation.

The intervention playbook

When an account trips the threshold, the moves are boring and effective:

• Executive check-in within 48 hours. Not a survey. A call from someone senior, referencing the specific friction.
• Clear the ticket queue before any renewal conversation. Unresolved tickets at renewal are a tax on every other argument you make.
• Re-onboard around the usage gap. Usage drops usually trace to one departed champion or one broken workflow. Fix the workflow, name a new champion.

The economics are lopsided. Saving that $48K account costs a few hours of senior time. Replacing it — at 2026’s 18-month CAC payback — costs roughly $70K of sales and marketing spend and a year and a half of waiting. Retention is the cheapest pipeline you own.

The same pipes run both directions

A health system tuned for risk also catches the happy inverse: usage up 60%, seats filling, a second team adopting. Those accounts are your expansion motion — detected by the same infrastructure, on the same alerts, with a different playbook attached.

This is the design behind ORL’s Retention Engine: product, support, and CRM signals wired into one account-health layer, with churn-risk and expansion alerts that arrive while there’s still time to act. The system doesn’t make the save. It makes the save possible — ninety days earlier than your renewal calendar would.