Retention and health signals

Beyond traditional health scores

Most customer health scores are built on a handful of metrics — login frequency, support ticket count, NPS rating. These capture part of the picture but miss critical context. A customer who logs in daily but never uses core features isn't healthy. A customer with zero tickets might be disengaged, not satisfied.

"Our health score said green. The customer said they were evaluating alternatives. The score had no idea." — Customer Success Manager

How Propane builds a richer health model

Propane combines product usage patterns, support interaction sentiment, billing history, and conversation context into a dynamic health assessment. The model adapts as it learns which signal combinations predict retention and which predict churn.

Leading indicators

Traditional health scores are lagging — they tell you a customer is unhealthy after the damage is done. Propane focuses on leading indicators: sentiment trajectory, usage pattern changes, and engagement shifts that precede churn by weeks.

Account-level context

Every health signal comes with context. You don't just see that an account's health dropped — you see the support ticket that triggered it, the feature they're missing, and the conversation where they expressed frustration.

Prioritized outreach

Not all at-risk accounts need the same response. Propane prioritizes based on revenue impact, contract timing, and the nature of the risk — so your team focuses on the accounts where intervention matters most.

"We stopped guessing which accounts to call first. The data tells us." — CS Operations Lead

Intervention recommendations

Based on the specific risk factors, Propane suggests the most effective intervention — a product walkthrough, a feature workaround, an executive check-in, or an escalation to engineering.

Who uses this

Customer success managers managing account portfolios. CS leaders allocating team resources. Revenue operations tracking retention and expansion metrics.