Asset performance, support volume, usage curves, contract terms, executive sentiment — fused into a health score that predicts churn 90+ days out, not 30.
5-dim composite health. Drift detection 90 days early. Expansion plays auto-drafted from the same model.
Health scores in CSPs lag 30-60 days. By the time the score turns red, the customer's already in market for a replacement. Health needs to fire 90+ days early to be useful.
The six capabilities that make this module work end-to-end. Pick any one as your starting point — they compound.
Asset uptime, usage trends, support volume, NPS, billing disputes, exec departures, public news. All fused into one score with provenance.
Every score change explains itself: which signals moved, which baselines they're scored against, what changed when. Not a black-box number.
Each risk pattern maps to a proven playbook — exec sponsor, technical deep-dive, contract restructure, hardware refresh. The right motion, not just an alert.
CSM, AE, exec sponsor capacity allocated to risk-weighted ARR — not ARR-band staffing.
Same model finds expansion candidates: green health + usage growth + new hires + recent renewal = ripe.
When a cohort's health drifts in aggregate — a region, a vertical, a product line — surface it. Systemic risk caught early.
28 signals flow in continuously, fuse into a per-customer score, surface playbooks, route capacity. CSMs spend their day on real risk — not on ARR-tier email cadences.
Asset, usage, support, billing, NPS, exec, public news — refreshed continuously.
ContinuousCausal model with per-account baselines. Each component scored separately and combined.
< 5minTop contributors surfaced. Drift vs cohort vs prior period. CSMs see the why immediately.
RealtimePattern matched to playbook. Exec sponsor assigned by ARR + risk pattern. Tasks fire.
< 30sDifferent businesses fail differently. Hardware-heavy customers signal churn through asset performance. SaaS-heavy through usage decay. Health weights tune to your actual churn drivers — calibrated against your historical churn cohort.
Re-calibrated quarterly as your business evolves. The model that called Q1 churns isn't the right one for Q3.
# Health Weights · calibrated against 38 churn events health "customer-base": # asset signals asset.uptime_30d weight 0.18 asset.anomalies_30d weight 0.12 # usage signals usage.trend_qoq weight 0.14 usage.feature_breadth weight 0.06 # support signals support.p1_count weight 0.10 support.escalations weight 0.08 # billing billing.dispute_rate weight 0.07 # NPS nps.detractor_count weight 0.09 # exec sentiment exec.departure_signal weight 0.16 thresholds: red < 50 # exec playbook yellow < 70 # csm playbook green >= 85 # expansion
Gainsight, Catalyst, ChurnZero — Health enriches your CSP, doesn't replace it.
Zendesk, Salesforce Service, Intercom — volume, escalation, resolution time.
Mixpanel, Amplitude, Pendo — feature breadth, login decay, depth.
Salesforce, HubSpot — renewal opps tagged with health and recommended motion.
We'll connect Health to your stack and surface the customers who are signaling churn 90+ days out — before your CSP shows red.