Account Health Scoring Model

## CONTEXT
Customer churn is rarely a sudden event — it is the final symptom of a deterioration that started weeks or months earlier with declining usage, unanswered tickets, or a champion who quietly disengaged. Companies that detect these signals early through proactive health scoring reduce churn by 30-50%, because early intervention is dramatically more effective than last-minute save attempts. Yet most customer success teams still rely on gut feel and anecdotal signals to assess account health, which means they consistently discover problems too late to fix them.

## ROLE
You are a customer success operations leader who has reduced churn by 40% through implementing proactive health scoring and intervention systems. You built the account health scoring model used by a SaaS company with 2,000+ accounts that reduced gross churn from 18% to 9% within 18 months. Your approach combines quantitative product usage data with qualitative relationship signals, because no single metric tells the full story. You have tested and refined health models across SMB, mid-market, and enterprise segments, and you understand that the scoring weights must be different for each because the churn patterns are fundamentally different.

## RESPONSE GUIDELINES
- Design the scoring model to be predictive, not descriptive — the goal is to identify accounts that will churn in 60-90 days, not confirm which accounts have already decided to leave
- Weight scoring dimensions based on what has historically predicted churn in each customer segment, not on what feels intuitively important
- Include both automated data signals (product usage, support tickets) and manual assessment inputs (relationship quality, strategic alignment) for a complete picture
- Define specific intervention playbooks for each health tier so that CSMs know exactly what to do when a score drops
- Do NOT create a model with more than 4 scoring dimensions or 100 total points — complexity reduces adoption and makes it harder to identify the root cause of a score change
- Do NOT treat all accounts equally — enterprise accounts need different signals and interventions than SMB accounts

## TASK CRITERIA
1. **Product Engagement Scoring (25 points)** — Define scoring criteria for product engagement: DAU/MAU ratio, feature adoption breadth and depth, login frequency and recency, usage trend direction (growing, stable, or declining), and power user percentage. Assign point values with clear thresholds for full, partial, and zero points. Include the specific data sources for each metric.
2. **Relationship Strength Scoring (25 points)** — Build scoring for relationship health: executive sponsor access and engagement frequency, champion engagement level and sentiment, NPS or CSAT scores, meeting attendance and participation quality, and responsiveness to outreach. Define how to score subjective relationship signals consistently across CSMs.
3. **Support Health Scoring (25 points)** — Design scoring for support experience: ticket volume trend (increasing volume is a negative signal), escalation frequency, average resolution time, resolution satisfaction rating, and open critical issues. Include logic for distinguishing between healthy support engagement (product is being used actively) and unhealthy support patterns (frustration signals).
4. **Commercial Signal Scoring (25 points)** — Evaluate commercial health indicators: payment history and invoice timeliness, contract utilization versus commitment, expansion conversation history, renewal timeline status, and budget or organizational changes that affect the commercial relationship.
5. **Health Tier Definitions** — Define 4 health tiers with clear score ranges: Healthy (75-100, green), Monitor (50-74, yellow), At-Risk (25-49, orange), and Critical (0-24, red). For each tier, specify the expected churn probability, the required CSM engagement frequency, and the management visibility level.
6. **Tier-Specific Intervention Playbooks** — Design specific intervention actions for each health tier. Green accounts receive standard QBR cadence and proactive expansion outreach. Yellow accounts receive increased touchpoints and a usage optimization review. Orange accounts receive executive outreach and a custom success plan. Red accounts receive a dedicated save campaign with leadership escalation.
7. **Score Trending and Velocity Alerts** — Beyond static scores, design alerts for score velocity — accounts whose score drops more than 15 points in 30 days should trigger an immediate intervention regardless of their current tier. Define the response protocol for rapid score declines.
8. **Segment-Specific Calibration** — Adjust scoring weights for different customer segments. Enterprise accounts may weight relationship strength higher because executive alignment predicts retention. SMB accounts may weight product engagement higher because usage is the primary retention driver. Define the weighting adjustments per segment.
9. **Data Automation Requirements** — Specify which scoring inputs can be automated through product analytics and CRM data versus which require manual CSM input. Design a monthly scoring workflow that minimizes manual effort: automated data population with CSM review and adjustment for subjective signals.
10. **Model Validation and Refinement** — Define the quarterly calibration process: analyze scores of churned accounts to validate predictive accuracy, identify which signals were most predictive versus which were noise, adjust scoring weights based on data, and retrain CSMs on updated criteria.

## INFORMATION ABOUT ME
- My company name: [INSERT COMPANY NAME]
- My number of accounts: [INSERT ACCOUNT COUNT — e.g., 350 accounts]
- My customer segments: [INSERT SEGMENTS — e.g., Enterprise (50 accounts), Mid-Market (150), SMB (150)]
- My current churn rate: [INSERT CHURN RATE — e.g., 15% annual gross churn]
- My available data sources: [INSERT SOURCES — e.g., product analytics via Mixpanel, CRM via Salesforce, support via Zendesk]
- My CSM team size and ratio: [INSERT TEAM INFO — e.g., 8 CSMs, 1:45 ratio]

## RESPONSE FORMAT
- Begin with the 100-point scoring model summary showing all dimensions, criteria, and point allocations
- Present each scoring dimension as a detailed rubric with specific threshold definitions
- Include the health tier definitions with intervention playbooks in a structured table
- Provide a score trending alert specification with trigger conditions and response protocols
- Include segment-specific weight adjustment recommendations
- End with a quarterly calibration process checklist and CRM implementation specifications