Catch churn your health score misses

Surface at-risk accounts using support, product, and billing data — before it's too late.

Every CX team has experienced it: a customer with a "green" health score suddenly cancels. The health score said they were fine. The reality was different.

The blind spots are structural. Health scores typically weight a few quantitative metrics — login count, ticket volume, survey score — and miss the qualitative signals that actually predict churn: tone shifts in support conversations, disengagement from key features, and unresolved frustrations that accumulate over months.

"We had a 94 health score. They churned the next week. I had no idea they were unhappy." — CSM, B2B SaaS

How Propane fills the gaps

Propane doesn't replace your health score — it enhances it with signals that traditional scoring misses. By analyzing the content and sentiment of every customer interaction, Propane catches the cases where the numbers look fine but the relationship is deteriorating.

Sentiment beneath the metrics

A customer who logs in every day but whose support tickets are increasingly frustrated is a churn risk that usage-based health scores miss entirely. Propane detects the sentiment trend and flags the disconnect.

Silent churn indicators

Some customers don't complain — they just leave. Propane identifies "silent churn" patterns: gradual disengagement, reduced feature breadth, fewer stakeholders logging in, and declining usage of sticky features.

Combining signals for accuracy

The power of Propane's approach is in combining signals that individually seem benign. A small drop in usage plus a slightly negative support interaction plus a skipped QBR — each is minor alone, but together they paint a concerning picture.

"Finally, something that tells me why an account is at risk, not just that it is." — VP Customer Success

Confidence scoring

Propane assigns a confidence level to each churn prediction, so your team can distinguish between "watch this account" and "intervene immediately." No more alert fatigue from oversensitive health scores.

Who uses this

CX teams with existing health scoring that want to catch the edge cases. CS leaders responsible for renewal rates. Revenue teams that need early warning on at-risk ARR.