Sales AI Skill
Health Score Churn Prevention
Calculate customer health scores from usage, engagement, and support data to identify at-risk accounts and trigger proactive retention interventions. Use when monitoring customer health, predicting churn risk, building retention playbooks, or creating proac...
Health Score & Churn Prevention
Predict and prevent revenue loss through data-driven health scoring and proactive intervention.
Workflow
- Ingest multi-dimensional customer data: usage, support tickets, NPS, billing, engagement.
- Calculate composite health score (0-100) using weighted model across all dimensions.
- Segment customers into health tiers: Healthy, At-Risk, Critical.
- Auto-trigger intervention workflows based on health tier and churn risk score.
- Assign at-risk accounts to CSM with prioritized action plan.
- Track intervention effectiveness and iterate on scoring model.
- Generate monthly churn risk report for leadership.
Health Score Model
HEALTH SCORE CALCULATION (0-100)
==================================
DIMENSION | WEIGHT | METRIC | Score Calculation
-------------------|--------|--------------------------------------|------------------
USAGE (30%) | 30 pts | |
| | Login frequency (last 30 days) | 0-5 pts
| | Feature adoption breadth | 0-5 pts
| | Core workflow completion rate | 0-10 pts
| | Usage trend ( WoW / MoM) | 0-5 pts
| | Storage/capacity utilization | 0-5 pts
| | Last active date | 0-5 pts
ENGAGEMENT (20%) | 20 pts | |
| | NPS/CSAT score | 0-5 pts
| | Meeting attendance rate | 0-5 pts
| | Email response rate | 0-5 pts
| | Training completion | 0-5 pts
SUPPORT (20%) | 20 pts | |
| | Open ticket count (critical) | 0-5 pts
| | Avg resolution time vs SLA | 0-5 pts
| | Ticket sentiment trend | 0-5 pts
| | Repeat issue frequency | 0-5 pts
FINANCIAL (15%) | 15 pts | |
| | Payment history (late payments) | 0-5 pts
| | Seat utilization rate | 0-5 pts
| | Expansion trend (upsell/cross-sell) | 0-5 pts
RELATIONSHIP (15%) | 15 pts | |
| | Stakeholder count (multi-threaded) | 0-5 pts
| | Executive sponsor engagement | 0-5 pts
| | Contract renewal date proximity | 0-5 pts
HEALTH TIERS:
80-100: GREEN — Healthy (low churn risk, focus on expansion)
60-79: YELLOW — Caution (moderate risk, proactive check-in)
40-59: ORANGE — At Risk (high risk, intervention required)
20-39: RED — Critical (immediate action, executive escalation)
0-19: BLACK — Imminent Churn (retention specialist + discount authority)
CHURN RISK PREDICTION:
Model outputs probability of churn within:
- 30 days: [%]
- 60 days: [%]
- 90 days: [%]
- 180 days: [%]
Based on historical patterns, usage decline trajectories,
and comparable customer churn events.
Intervention Playbook by Health Tier
INTERVENTION PLAYBOOK
======================
GREEN (80-100) — NURTURE & EXPAND:
────────────────────────────────
Action Items:
1. Schedule QBR/EBR (if not already on cadence)
2. Identify expansion opportunities (additional seats, modules, features)
3. Request case study or reference commitment
4. Invite to customer advisory board
5. Send quarterly value report (ROI achieved)
Trigger: Health score >80
Owner: Account Manager / CSM
Timeline: Standard quarterly cadence
Budget: Standard CSM allocation
YELLOW (60-79) — PROACTIVE CHECK-IN:
──────────────────────────────────
Action Items:
1. CSM scheduled check-in call within 5 business days
2. Review usage patterns and identify drop-off areas
3. Offer targeted training or onboarding refresher
4. Check for unresolved support issues
5. Survey stakeholder satisfaction
6. Document risk factors and monitor weekly
Trigger: Health score drops below 80 OR single dimension drops >20%
Owner: CSM
Timeline: Within 5 business days
Budget: Standard CSM allocation + training resources
ORANGE (40-59) — ACTIVE RETENTION:
─────────────────────────────────
Action Items:
1. Executive check-in within 48 hours (VP CSM or Director)
2. Root cause analysis: why did health decline?
3. Create 30-day turnaround plan with customer
4. Assign dedicated success resources (if applicable)
5. Review and resolve ALL open support tickets
6. Executive-to-executive outreach (CEO to customer exec)
7. Weekly health score monitoring
8. Consider retention offer (extended term, bonus features)
Trigger: Health score drops below 60 OR critical dimension fails
Owner: Senior CSM + VP Customer Success
Timeline: Immediate (48 hours)
Budget: Retention budget allocated (up to 15% discount authority)
RED (20-39) — CRITICAL RETENTION:
────────────────────────────────
Action Items:
1. War room: cross-functional retention team assembled
2. Customer executive call within 24 hours
3. Full health diagnostic + customer sentiment deep-dive
4. Custom retention offer prepared (discount, credits, enhanced SLA)
5. Alternative solutions discussed (downgrade vs. churn)
6. Daily health score monitoring
7. Escalation to CRO/CFO if deal value exceeds threshold
Trigger: Health score drops below 40 OR explicit churn signals
Owner: Director of Customer Success + Legal + Finance
Timeline: Immediate (24 hours)
Budget: Up to 30% discount + custom terms (requires CFO approval)
BLACK (0-19) — IMMINENT CHURN:
────────────────────────────
Action Items:
1. CEO-level outreach (personal call or visit)
2. Maximum retention offer authorized
3. Post-mortem analysis: what went wrong and why
4. If churn confirmed: graceful offboarding + winback plan
5. Lessons documented and shared with product/sales teams
6. Re-engagement campaign scheduled (90/180/365 days)
Trigger: Health score <20 OR contract termination notice received
Owner: VP/Head of Customer Success + CEO (for strategic accounts)
Timeline: Immediate
Budget: Full retention authority (requires executive approval)
Churn Signal Detection
AUTOMATED CHURN SIGNALS
========================
HIGH SEVERITY SIGNALS (Immediate Alert):
─────────────────────────────────────
- NPS score drops from 8+ to ≤4
- All user logins stop for 14+ days
- Contract termination request received
- Customer explicitly mentions competitor
- Legal complaint or formal notice
- Payment failure with no communication
- Key champion leaves company
MEDIUM SEVERITY SIGNALS (48-Hour Review):
──────────────────────────────────────
- Usage drops >40% week-over-week
- Support ticket volume spikes >3x normal
- NPS/CSAT drops 2+ points from average
- QBR/EBR meeting declined 2+ times
- Email response rate drops below 20%
- Seat count decreasing (users offboarding)
- Feature adoption declining (new features unused)
LOW SEVERITY SIGNALS (Weekly Monitoring):
──────────────────────────────────────
- Usage trend declining gradually (>10% MoM)
- Meeting attendance inconsistent
- Single support ticket with negative sentiment
- Email open rate declining
- One stakeholder disengaged
- Payment slightly late (1-5 days)
SIGNAL COMBINATION EFFECT:
- 1 low signal: Monitor
- 2+ low signals: Flag for CSM review
- 1 medium + 1 low: Trigger Yellow intervention
- 2+ medium signals: Trigger Orange intervention
- 1 high signal: Trigger Red intervention
- Any high signal: Immediate escalation regardless of score
Churn Analytics & Reporting
CHURN ANALYTICS DASHBOARD
============================
Portfolio Health Overview:
Total Customers: [count]
Healthy (Green): [count] ([%])
Caution (Yellow): [count] ([%])
At Risk (Orange): [count] ([%])
Critical (Red/Black): [count] ([%])
Average Health Score: [X/100]
Churn Metrics:
Monthly Churn Rate: [%] (target: <2%)
Annualized Revenue Churn: [%] (target: <10%)
Gross Revenue Retention (GRR): [%] (target: >90%)
Net Revenue Retention (NRR): [%] (target: >110%)
Churned Customers This Month: [count]
Revenue Lost from Churn: [$ amount]
Revenue Saved from Retention: [$ amount]
Saved-to-Lost Ratio: [X:1]
Intervention Effectiveness:
Interventions triggered this month: [count]
Intervention success rate: [%] (health improved post-intervention)
Average time to recovery (Orange → Green): [X days]
Average time to recovery (Red → Yellow): [X days]
Retention offer acceptance rate: [%]
Average discount on retention offers: [%]
Churn Root Cause Analysis:
Top Churn Reasons (This Quarter):
1. [Reason] — [count] ([%])
2. [Reason] — [count] ([%])
3. [Reason] — [count] ([%])
4. [Reason] — [count] ([%])
5. [Reason] — [count] ([%])
Churn by Customer Segment:
SMB: [%] | Mid-Market: [%] | Enterprise: [%]
Churn by Product Tier:
Basic: [%] | Pro: [%] | Enterprise: [%]
Churn by Tenure:
<6 months: [%] | 6-12 months: [%] | 1-2 years: [%] | 2+ years: [%]
Predictive Model Performance:
Churn prediction accuracy: [%] (true positive rate)
False positive rate: [%] (healthy customers flagged)
False negative rate: [%] (churned customers missed)
Model precision: [%] | Model recall: [%]
Model F1 score: [X]
Edge Cases
- New customer (<30 days): Insufficient data for reliable health score
- Mitigation: Use onboarding completion score instead (tasks completed, features adopted)
- Weight engagement and relationship dimensions higher until usage data accumulates
- Set lower intervention thresholds during first 90 days
- Assign dedicated onboarding specialist regardless of score
- Seasonal usage patterns: Customer with predictable seasonal dips (e.g., retail, education)
- Mitigation: Compare usage to same period last year (YoY) instead of rolling average
- Tag account with seasonal profile; adjust scoring model accordingly
- Suppress false alerts during known low-usage periods
- Flag when seasonal pattern breaks (usage below seasonal baseline)
- Enterprise account with many sub-accounts: Health varies significantly across divisions
- Mitigation: Calculate health at both parent and subsidiary level
- Parent score = weighted average based on revenue contribution
- Flag individual subsidiary health even if parent score looks healthy
- Assign separate CSM owners per major division
- Churn signal false positives: Customer flagged as at-risk but actually in active implementation
- Mitigation: Cross-reference with implementation phase; suppress alerts during active rollout
- Check if support tickets are "implementation questions" vs. "problem complaints"
- Include implementation status as context in health score display
- CSM manual override to temporarily suspend automated alerts
Integration Points
- Salesforce/HubSpot: Customer records, contract dates, opportunity history
- Intercom/Gainsight: Customer engagement data, messaging history
- Zendesk/Freshdesk: Support ticket volume, resolution times, sentiment
- Product analytics (Mixpanel/Amplitude): Usage data, feature adoption, login frequency
- Billing (Stripe/Chargebee): Payment history, MRR, expansion revenue
- NPS/CSAT (Delighted/SurveyMonkey): Satisfaction scores and trends
- Slack/Teams: Alert delivery to CSM team channels
- Gong: Call sentiment and engagement analysis for relationship scoring