| name | customer-health-score |
| description | Designs customer health scoring models with engagement metrics, risk indicators, and intervention triggers to prevent churn proactively. |
| allowed-tools | Read Write Glob |
| metadata | {"author":"matthewhitcham","version":"1.0"} |
Customer Health Score
When to Use This Skill
Use this skill when you need to:
- Build a scoring model that identifies at-risk customers before they churn
- Define engagement metrics and risk indicators for customer accounts
- Create intervention triggers and playbooks for different health levels
- Prioritize customer success efforts based on objective health data
DO NOT use this skill for NPS surveys, customer segmentation by value, or sales pipeline scoring. This is for ongoing customer health monitoring.
Core Principle
CUSTOMER HEALTH IS A LEADING INDICATOR — BY THE TIME A CUSTOMER TELLS YOU THEY ARE LEAVING, THE DECISION WAS MADE WEEKS AGO. A HEALTH SCORE GIVES YOU THOSE WEEKS BACK.
Phase 1: Define Health Signals
Identify what healthy and unhealthy customers look like.
Required Inputs
| Input | What to Ask | Default |
|---|
| Business model | "Subscription, project-based, retainer, or product?" | Subscription |
| Healthy customer profile | "Describe your most engaged, happiest customer. What do they do?" | No default |
| Churn signals | "Think of customers who left. What did they do (or stop doing) before leaving?" | No default |
| Data available | "What customer data do you track? (login frequency, usage, support tickets, payment history)" | Basic — email engagement, payment |
| Customer count | "How many active customers do you have?" | Under 50 |
GATE: Confirm health signals before building the scoring model.
Phase 2: Build Scoring Model
Create a weighted health score from 0-100.
Health Score Components
## Customer Health Score Model
**Score range:** 0-100
**Healthy:** 70-100
**At risk:** 40-69
**Critical:** 0-39
### Scoring Components
| Component | Weight | Metric | Scoring |
|-----------|--------|--------|---------|
| Engagement | 30% | [Usage frequency, login rate, feature adoption] | 0-30 points |
| Relationship | 25% | [Communication responsiveness, meeting attendance, feedback participation] | 0-25 points |
| Financial | 25% | [Payment timeliness, expansion revenue, contract length] | 0-25 points |
| Satisfaction | 20% | [NPS, CSAT, support ticket sentiment] | 0-20 points |
Component Scoring Detail
For each component, define the scoring tiers:
### Engagement Score (0-30 points)
| Behavior | Points | Rationale |
|----------|--------|-----------|
| Active weekly (uses product/service regularly) | 30 | Fully engaged |
| Active biweekly | 20 | Moderately engaged |
| Active monthly | 10 | Low engagement — intervention needed |
| Inactive 30+ days | 0 | Critical — immediate outreach |
### Relationship Score (0-25 points)
| Behavior | Points |
|----------|--------|
| Responds to outreach within 48 hours | 10 |
| Attends scheduled meetings/calls | 10 |
| Provides feedback when asked | 5 |
### Financial Score (0-25 points)
| Behavior | Points |
|----------|--------|
| Pays on time consistently | 15 |
| Has expanded (upsold, renewed, added services) | 10 |
| Late payments or disputes | -10 |
### Satisfaction Score (0-20 points)
| Signal | Points |
|--------|--------|
| NPS 9-10 (Promoter) | 20 |
| NPS 7-8 (Passive) | 10 |
| NPS 0-6 (Detractor) | 0 |
| Support complaints increasing | -5 |
GATE: Present scoring model for validation.
Phase 3: Intervention Playbooks
Define actions for each health zone.
Intervention Matrix
## Health Zone Actions
### Healthy (70-100) — Nurture and Expand
**Check-in frequency:** Monthly
**Actions:**
- Send value-add content and tips
- Identify upsell or referral opportunities
- Request testimonials or case studies
- Invite to advisory board or beta programs
### At Risk (40-69) — Engage and Recover
**Check-in frequency:** Biweekly
**Actions:**
- Personal outreach from account owner (not automated)
- Identify specific decline — which component dropped?
- Offer training, support, or strategy session
- Address any unresolved issues immediately
### Critical (0-39) — Save or Prepare
**Check-in frequency:** Weekly
**Actions:**
- Escalate to business owner
- Direct phone call or video meeting (not email)
- Ask explicitly: "What would need to change for this to work?"
- Offer concession if appropriate (discount, extended trial, service credit)
- If unrecoverable, conduct exit interview for learning
Alert System
## Health Score Alerts
| Trigger | Alert | Action |
|---------|-------|--------|
| Score drops below 70 | Email to account owner | Schedule check-in within 5 days |
| Score drops below 40 | Email to business owner | Call customer within 48 hours |
| Score drops 20+ points in one period | Urgent flag | Same-day outreach |
| Score increases above 80 | Opportunity flag | Send referral or upsell prompt |
Phase 4: Operationalize
Make health scoring part of the regular business rhythm.
Scoring Cadence
- Monthly: Update all customer health scores
- Weekly: Review any customers flagged as Critical
- Quarterly: Analyze trends — is overall health improving or declining?
Health Dashboard
## Customer Health Overview — [Month]
| Health Zone | Count | % of Total | Revenue at Risk |
|-------------|-------|-----------|----------------|
| Healthy (70-100) | [#] | [%] | — |
| At Risk (40-69) | [#] | [%] | $[X] |
| Critical (0-39) | [#] | [%] | $[X] |
**Average health score:** [X]
**Trend:** Improving / Stable / Declining
Quarterly Calibration
Review and adjust the model quarterly:
- Did health scores predict actual churn?
- Are the component weights right?
- Are any metrics too hard to collect consistently?
- Do the intervention playbooks work?
Anti-Patterns
- Score without action — a health score is useless if no one acts on it. Every zone needs a defined playbook.
- Over-complicated model — start with 3-4 components. Adding 10 metrics makes the score hard to maintain and interpret.
- Scoring only on gut feel — the score must be based on observable, measurable behaviors, not "I think they are happy."
- Ignoring healthy customers — high health scores need attention too. Neglected promoters become passives.
- Static model — customer behavior patterns change. Recalibrate the model quarterly.
Recovery
- User does not track enough data: Start with what is available — even payment history and email response rate create a useful 2-component score.
- Too few customers to score: Under 10 customers, skip the score and maintain a simple traffic-light status (green/yellow/red) with monthly check-ins.
- Health score does not predict churn: The wrong metrics are being weighted. Analyze actual churned customers and reverse-engineer which signals mattered.
- User cannot act on scores (no time): Automate healthy-zone nurturing (drip emails, tips). Spend manual time only on at-risk and critical accounts.
- All customers show as healthy but churn happens: The scoring thresholds are too generous. Lower the "healthy" threshold and add more sensitive indicators.
