Customer Health Score
A composite index that blends usage, support, sentiment, and engagement signals into one number meant to predict whether an account will renew or churn.
Health score = Σ(signal score × weight), normalized to a 0-100 scale or red/yellow/green bandA customer health score rolls several signals — product usage, support activity, sentiment, account engagement, sometimes payment and relationship depth — into a single index, usually a 0-100 number or a red/yellow/green band. It is a customer-success construct, but support feeds it heavily: ticket volume, escalation frequency, reopens, and the tone of recent conversations are among the earliest signs an account is slipping, often visible before usage sags.
There is no standard formula. Each team picks a handful of inputs and assigns weights it believes track retention — usage might count for 40%, support trends 25%, sentiment 20%, relationship depth 15% — then normalizes to a common scale. The weights are the whole game, and most are set by gut rather than fitted against actual churn, then rarely revisited as the product and customer base change underneath them.
What it hides: a health score is only as predictive as the data and weighting behind it, and surveys of customer-success leaders repeatedly find that most don't trust their score to reliably flag churn. The usual culprit is flat thresholds: an account that dropped from daily to weekly logins reads the same green as one that was always weekly, even though one is quietly leaving. A green dashboard can mean a healthy base or a stale model, and the number itself rarely tells you which — so a score that isn't regularly checked against who actually churned is decoration, not a forecast.