Customer Health Analyst

Customer Health Analyst

Expert guidance for customer health scoring, predictive analytics, and data-driven customer success strategies. Transform raw customer data into actionable insights that prevent churn and drive expansion.

Philosophy

Customer health is not a single metric — it's a predictive system:

1. Measure what matters — Health scores should predict outcomes, not just track activity 2. Lead, don't lag — Focus on indicators that predict churn before it's too late 3. Segment for action — Different customers need different interventions 4. Automate detection — Scale health monitoring across your entire customer base 5. Close the loop — Analytics without action is just expensive data collection

How This Skill Works

When invoked, apply the guidelines in rules/ organized by:

• health-* — Health score design, weighting, and calibration
• indicators-* — Leading vs lagging indicator analysis
• churn-* — Prediction modeling and early warning systems
• usage-* — Analytics and adoption metrics
• risk-* — Identification, escalation, and intervention
• data-* — Enrichment and customer 360 development
• cohort-* — Analysis and benchmarking
• executive-* — Reporting and dashboards
• segmentation-* — Customer tiers and scoring models

Core Frameworks

The Health Score Hierarchy

┌─────────────────────────────────────────────────────────────────┐
│                    COMPOSITE HEALTH SCORE                       │
│                         (0-100)                                 │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌──────────┐       │
│  │ PRODUCT  │  │ENGAGEMENT│  │ GROWTH   │  │ SUPPORT  │       │
│  │  USAGE   │  │          │  │ SIGNALS  │  │ HEALTH   │       │
│  │  (35%)   │  │  (25%)   │  │  (20%)   │  │  (20%)   │       │
│  └──────────┘  └──────────┘  └──────────┘  └──────────┘       │
│                                                                 │
├─────────────────────────────────────────────────────────────────┤
│                    COMPONENT METRICS                            │
│                                                                 │
│  Usage:        Engagement:    Growth:        Support:          │
│  - DAU/MAU     - NPS score    - Seat trend   - Ticket volume   │
│  - Features    - CSM meetings - Usage trend  - Resolution time │
│  - Depth       - Email opens  - Expansion    - Sentiment       │
│  - Breadth     - Logins       - Contract     - Escalations     │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Leading vs Lagging Indicators

Customer Health States

┌─────────────────────────────────────────────────────────────────┐
│                                                                 │
│  THRIVING ──→ HEALTHY ──→ NEUTRAL ──→ AT-RISK ──→ CRITICAL    │
│    (85+)      (70-84)     (50-69)     (30-49)      (<30)       │
│                                                                 │
│  Expand       Monitor     Engage      Intervene    Escalate    │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Health Score Components

Churn Risk Factors

The Analytics Stack

Key Metrics

Executive Dashboard KPIs

Cohort Analysis Framework

Anti-Patterns

• Vanity health scores — Scores that look good but don't predict outcomes
• Over-weighted product usage — Ignoring relationship and sentiment signals
• Lagging indicator focus — Measuring what already happened
• One-size-fits-all thresholds — Same scores mean different things for different segments
• Manual-only health tracking — Can't scale without automation
• Score without action — Calculating risk without intervention playbooks
• Annual calibration only — Health models need continuous refinement
• Ignoring data quality — Garbage in, garbage out

1. Health Score Design (health)

Impact: CRITICAL Description: Health score architecture, component selection, weight assignment, scoring algorithms, threshold calibration, and model validation.

2. Leading vs Lagging Indicators (indicators)

Impact: CRITICAL Description: Indicator identification, predictive signal development, correlation analysis, signal prioritization, and action trigger design.

3. Churn Prediction (churn)

Impact: CRITICAL Description: Prediction model development, feature engineering, risk scoring, early warning systems, and intervention timing optimization.

4. Usage Analytics (usage)

Impact: HIGH Description: Engagement measurement, feature adoption tracking, usage patterns, behavioral analysis, and adoption benchmarking.

5. Risk Identification (risk)

Impact: CRITICAL Description: Risk signal detection, escalation frameworks, intervention playbooks, stakeholder communication, and save strategies.

6. Data Enrichment (data)

Impact: HIGH Description: Data source integration, enrichment strategies, data quality management, 360-degree customer view, and data governance.

7. Cohort Analysis (cohort)

Impact: HIGH Description: Cohort definition, retention curve analysis, comparative benchmarking, segment performance, and trend identification.

8. Executive Reporting (executive)

Impact: HIGH Description: KPI selection, dashboard design, data storytelling, executive presentations, and board reporting.

9. Segmentation & Scoring (segmentation)

Impact: MEDIUM-HIGH Description: Customer tier definition, behavioral clustering, value-based segmentation, scoring model design, and segment-specific strategies.

Churn Prediction Modeling

Impact: CRITICAL

Effective churn prediction gives you 60-90 days of lead time to intervene. A well-calibrated model can reduce churn by 15-30% by enabling proactive outreach to at-risk accounts before they decide to leave.

The Churn Prediction Pipeline

┌──────────────────────────────────────────────────────────────────┐
│                    CHURN PREDICTION PIPELINE                     │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│  DATA           FEATURES         MODEL           SCORING        │
│  COLLECTION  →  ENGINEERING  →   TRAINING   →   & ALERTS       │
│                                                                  │
│  • Product       • Usage decay    • Logistic     • Daily risk   │
│  • CRM           • Engagement     • Random       • Threshold    │
│  • Support       • Sentiment      • XGBoost      • Routing      │
│  • Financial     • Growth         • Neural       • Actions      │
│                                                                  │
├──────────────────────────────────────────────────────────────────┤
│                    FEEDBACK LOOP                                 │
│                                                                  │
│           Actual Outcomes → Model Refinement → Improved Accuracy │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Feature Categories for Churn Models

Good Feature Engineering

Feature: Usage Velocity (30-Day)

Definition:
velocity_30d = (usage_current - usage_30d_ago) / usage_30d_ago

Why It's Predictive:
- Captures direction AND magnitude of change
- Declining velocity precedes churn by 60-90 days
- More predictive than static usage levels

Implementation:
SELECT
  customer_id,
  (current_usage - lag_30d_usage) / NULLIF(lag_30d_usage, 0) as velocity_30d
FROM customer_usage
WHERE lag_30d_usage > 0

Feature Distribution:
- Retained customers: mean velocity = +0.05
- Churned customers: mean velocity = -0.28
- Separation is clear and actionable

Bad Feature Engineering

Feature: Total Logins (All-Time)

Problems:
✗ Doesn't account for tenure
✗ No directional information
✗ Old customers always score higher
✗ Not predictive of future behavior

Better Alternative:
- Login frequency (logins per week)
- Login trend (this month vs. last month)
- Days since last login

Feature Reality:
- Retained customers: mean = 1,247 logins
- Churned customers: mean = 892 logins
- Overlap is massive, low predictive value

Model Selection Guide

Model Training Process

Step 1: Data Preparation
├── Define churn (90-day non-renewal? Contract cancellation?)
├── Set observation window (features from T-90 to T-0)
├── Set outcome window (churn in next 90 days)
└── Handle class imbalance (SMOTE, class weights)

Step 2: Feature Selection
├── Calculate feature importance (univariate)
├── Remove correlated features (>0.8 correlation)
├── Engineer interaction features
└── Normalize/standardize as needed

Step 3: Model Training
├── Split: 70% train, 15% validation, 15% test
├── Train multiple model types
├── Tune hyperparameters on validation set
└── Select best model by validation AUC

Step 4: Model Evaluation
├── Test set performance (AUC, precision, recall)
├── Calibration check (predicted vs. actual probabilities)
├── Feature importance review
└── Business metric simulation

Step 5: Deployment
├── Productionize scoring pipeline
├── Set up monitoring and alerts
├── Document model and features
└── Plan retraining schedule

Model Performance Metrics

Threshold Selection

Tradeoff: Precision vs. Recall

High Threshold (e.g., >0.7 probability):
✓ High precision (fewer false positives)
✗ Low recall (miss some actual churns)
→ Use when intervention is expensive

Low Threshold (e.g., >0.3 probability):
✓ High recall (catch more actual churns)
✗ Low precision (more false positives)
→ Use when missing churn is expensive

Optimal Threshold:
- Calculate cost of false positive (unnecessary intervention)
- Calculate cost of false negative (missed churn)
- Find threshold that minimizes total expected cost

Risk Tiering System

Early Warning System Design

┌──────────────────────────────────────────────────────────────────┐
│                    EARLY WARNING SYSTEM                          │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│  Daily Scoring Pipeline:                                         │
│  ├── Pull latest customer data                                   │
│  ├── Calculate features                                          │
│  ├── Score all customers                                         │
│  └── Update risk tiers                                           │
│                                                                  │
│  Alert Triggers:                                                 │
│  ├── Risk tier change (e.g., Low → Medium)                      │
│  ├── Probability increase >20 points                             │
│  ├── Critical signals detected                                   │
│  └── Combination triggers                                        │
│                                                                  │
│  Alert Routing:                                                  │
│  ├── Critical → CSM + Manager + VP (Slack + Email)              │
│  ├── High → CSM + Manager (Email)                               │
│  ├── Medium → CSM (Dashboard + Email)                           │
│  └── Low → Dashboard only                                        │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Intervention Optimization

Model Monitoring Dashboard

Churn Prediction Model Health

┌─────────────────────────────────────────────────────────────┐
│  MODEL PERFORMANCE (Rolling 90 Days)                        │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  AUC-ROC:          0.78 (target: >0.75)        ✓           │
│  Precision:        0.62 (target: >0.60)        ✓           │
│  Recall:           0.71 (target: >0.70)        ✓           │
│  Lift at 10%:      4.2x (target: >3x)          ✓           │
│                                                             │
├─────────────────────────────────────────────────────────────┤
│  PREDICTION ACCURACY                                        │
│                                                             │
│  Actual Churns:         47                                  │
│  Predicted (>50%):      38                                  │
│  Correctly Predicted:   33                                  │
│  Surprise Churns:       14                                  │
│  False Alarms:          5                                   │
│                                                             │
├─────────────────────────────────────────────────────────────┤
│  FEATURE IMPORTANCE (Top 5)                                 │
│                                                             │
│  1. Usage velocity (30d)      ████████████  28%            │
│  2. NPS trend                 ████████      19%            │
│  3. Support sentiment         ███████       15%            │
│  4. Champion engagement       ██████        13%            │
│  5. Feature adoption trend    █████         11%            │
│                                                             │
├─────────────────────────────────────────────────────────────┤
│  ALERTS                                                     │
│                                                             │
│  ⚠ Recall dropped 5% vs. prior period                      │
│  ⚠ Feature drift detected in usage metrics                 │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Model Maintenance Schedule

Churn Model Checklist

□ Data Quality
  □ Churn definition is clear and consistent
  □ Historical data covers 12+ months
  □ Feature data is complete and accurate
  □ Class imbalance addressed appropriately

□ Feature Engineering
  □ Features are predictive (tested)
  □ No data leakage (future info in features)
  □ Features are interpretable
  □ Trends included, not just levels

□ Model Development
  □ Train/validation/test split done properly
  □ Cross-validation used for tuning
  □ Multiple model types compared
  □ Hyperparameters optimized

□ Model Evaluation
  □ Performance meets targets
  □ Model is calibrated (probabilities accurate)
  □ No obvious bias by segment
  □ Business simulation validates value

□ Deployment
  □ Scoring pipeline automated
  □ Monitoring in place
  □ Alerts configured
  □ Documentation complete

□ Operations
  □ Retraining schedule defined
  □ Drift monitoring active
  □ Feedback loop from CS team
  □ Regular accuracy reviews

Anti-Patterns

• Predicting the past — Data leakage giving false accuracy
• One model fits all — Ignoring segment differences
• Set and forget — Models decay without retraining
• Ignoring false positives — Intervention fatigue from bad predictions
• Probability as certainty — Treating 60% risk as definite churn
• No action mapping — Predictions without intervention playbooks
• Over-engineering — Complex models when simple works
• Ignoring surprise churns — Not investigating model failures

Cohort Analysis & Benchmarking

Impact: HIGH

Cohort analysis reveals patterns hidden in aggregate data. By grouping customers with shared characteristics and tracking them over time, you can identify which customer segments thrive, which struggle, and what drives the difference. Benchmarking puts your performance in context.

The Cohort Analysis Framework

┌──────────────────────────────────────────────────────────────────┐
│                    COHORT ANALYSIS PROCESS                       │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│  DEFINE          TRACK           ANALYZE         ACTION         │
│  COHORTS    →    OVER TIME   →   PATTERNS   →   INSIGHTS       │
│                                                                  │
│  • Time-based    • Retention     • Compare       • Why differ?  │
│  • Behavioral    • Revenue       • Identify      • What works?  │
│  • Value-based   • Engagement    • Benchmark     • Optimize     │
│  • Acquisition   • Health        • Trend         • Predict      │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Cohort Definition Types

Retention Cohort Analysis (Time-Based)

Monthly Retention by Signup Cohort

Cohort    Month 0   Month 1   Month 2   Month 3   Month 6   Month 12
────────────────────────────────────────────────────────────────────
Jan 2024    100%      88%       82%       78%       71%       65%
Feb 2024    100%      91%       85%       81%       74%       -
Mar 2024    100%      89%       84%       80%       72%       -
Apr 2024    100%      92%       87%       83%       -         -
May 2024    100%      90%       86%       -         -         -
Jun 2024    100%      93%       -         -         -         -
Jul 2024    100%      -         -         -         -         -

Insights:
✓ Month 1 retention improving (88% → 93%)
✓ Month 6 retention stable around 72%
⚠ Q1 cohorts showing lower long-term retention
Action: Investigate Jan cohort for onboarding issues

Revenue Retention Cohort Analysis

Net Revenue Retention by Signup Quarter

Cohort    Q0      Q1      Q2      Q3      Q4      Q5      Q6
────────────────────────────────────────────────────────────
Q1 2023   100%    98%    102%    108%    115%    118%    122%
Q2 2023   100%   101%    106%    112%    119%    124%     -
Q3 2023   100%    99%    104%    109%    116%     -       -
Q4 2023   100%   102%    108%    114%     -       -       -
Q1 2024   100%   103%    110%     -       -       -       -
Q2 2024   100%   104%     -       -       -       -       -

Analysis:
├── All cohorts achieve >100% NRR (expansion > churn)
├── Q2 2024 showing strongest early expansion
├── Typical trajectory: 100% → 110% → 120% by Year 2
└── Cohort maturity required for full picture

Good Cohort Visualization

Retention Curve by Customer Segment

100% ┤● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
     │  ╲
 90% ┤   ╲ ● ● ● ● ● ● ● ● ● ● ● ● ● Enterprise
     │     ╲ ╲
 80% ┤        ╲ ● ● ● ● ● ● ● ● ● ● Mid-Market
     │          ╲ ╲
 70% ┤             ╲ ● ● ● ● ● ● ● SMB
     │               ╲
 60% ┤                 ╲ ● ● ● ● Startup
     │
 50% ┤
     └────┬────┬────┬────┬────┬────┬────┬────┬────
          1    2    3    4    5    6    9   12
                    Months Since Signup

Key Insights:
1. Enterprise: 95% retention at month 12 (target: 90%)
2. Mid-Market: 82% retention at month 12 (on target)
3. SMB: 71% retention at month 12 (below 75% target)
4. Startup: 58% retention at month 12 (investigate)

Bad Cohort Analysis

Customer Retention Report

Total customers: 2,500
Active customers: 2,150
Retention rate: 86%

Problems:
✗ No time dimension
✗ No segmentation
✗ No trend analysis
✗ No benchmark comparison
✗ Point-in-time snapshot only
✗ Blends all cohort maturities
✗ No actionable insights

Behavioral Cohort Analysis

Retention by Activation Behavior (First 30 Days)

Behavior Cohort                    Month 6 Retention    Index
──────────────────────────────────────────────────────────────
Completed core workflow             89%                 1.48x
Invited 3+ team members             84%                 1.40x
Used 5+ features                    81%                 1.35x
Attended onboarding webinar         78%                 1.30x
Created 10+ [objects]               75%                 1.25x
Basic activation only               60%                 1.00x
No activation (signed up only)      32%                 0.53x

Implications:
1. Core workflow completion is strongest retention predictor
2. Team invitation = social commitment = retention
3. Focus onboarding on these high-impact behaviors
4. Users who don't activate are unlikely to retain

Value-Based Cohort Analysis

Retention & NRR by Initial ARR Tier

Tier          ARR Range        Logo Retention  NRR    Avg Health
─────────────────────────────────────────────────────────────────
Enterprise    >$100K           96%             135%   82
Upper MM      $50K-$100K       93%             122%   76
Lower MM      $20K-$50K        88%             112%   71
SMB           $5K-$20K         78%             98%    64
Startup       <$5K             62%             85%    52

Insights:
├── Enterprise segment is profitable (high retention, expansion)
├── SMB requires efficiency focus (lower retention, no expansion)
├── Startup segment may not be viable at scale
├── Health score correlates with retention across tiers
└── Consider minimum viable customer criteria

Benchmarking Framework

Industry Benchmark Sources

Cohort Comparison Best Practices

Comparing Cohorts Effectively

1. Same Time Window
   ✓ Compare Jan 2024 at Month 6 to Jan 2023 at Month 6
   ✗ Compare Jan 2024 at Month 6 to Jan 2023 at Month 12

2. Normalize for Seasonality
   ✓ Account for holiday slowdowns, fiscal year patterns
   ✗ Compare Q4 directly to Q1 without adjustment

3. Statistical Significance
   ✓ Ensure cohort size supports conclusions (n > 30)
   ✗ Draw conclusions from cohorts of 5 customers

4. Consistent Definitions
   ✓ Same retention definition across cohorts
   ✗ Changing what "active" means mid-analysis

5. Account for Mix Shifts
   ✓ Note if segment composition changed
   ✗ Compare blended metrics when mix shifted significantly

Cohort Analysis Dashboard

Cohort Analysis Dashboard

┌─────────────────────────────────────────────────────────────────┐
│  RETENTION TRENDS                                               │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  Month 1 Retention (6-month trend):   91% → 89% → 90% → 93%   │
│  Status: ✓ Improving                                            │
│                                                                 │
│  Month 6 Retention (6-month trend):   72% → 71% → 73% → 74%   │
│  Status: ✓ Stable/Improving                                     │
│                                                                 │
│  Month 12 Retention (trailing):       65%                       │
│  Status: ⚠ Below 70% target                                    │
│                                                                 │
├─────────────────────────────────────────────────────────────────┤
│  SEGMENT COMPARISON (Month 6)                                   │
│                                                                 │
│  Enterprise:  ████████████████████ 94%  (↑ vs prior)           │
│  Mid-Market:  █████████████████░░░ 82%  (= vs prior)           │
│  SMB:         ██████████████░░░░░░ 71%  (↓ vs prior)           │
│  Startup:     ████████████░░░░░░░░ 58%  (↓ vs prior)           │
│                                                                 │
│  ⚠ Alert: SMB retention declining - investigate                │
│                                                                 │
├─────────────────────────────────────────────────────────────────┤
│  BEHAVIORAL COHORT INSIGHTS                                     │
│                                                                 │
│  Highest retention cohort: Multi-user activation (89%)         │
│  Lowest retention cohort: Single feature users (52%)           │
│  Biggest gap: 37 percentage points                              │
│                                                                 │
│  Recommendation: Focus onboarding on multi-user + multi-feature │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Cohort Analysis Checklist

□ Cohort Definition
  □ Clear criteria for cohort membership
  □ Mutually exclusive cohorts (no overlap)
  □ Meaningful segment differences
  □ Sufficient sample size per cohort

□ Metric Selection
  □ Primary metric defined (retention, NRR, etc.)
  □ Time windows specified
  □ Calculation methodology documented
  □ Edge cases handled (partial periods, etc.)

□ Data Preparation
  □ Data completeness verified
  □ Historical data sufficient for trends
  □ Consistent definitions over time
  □ Cohort assignment logic validated

□ Analysis Execution
  □ Retention curves plotted
  □ Segment comparisons completed
  □ Trends over time identified
  □ Statistical significance checked

□ Benchmarking
  □ Internal benchmarks established
  □ Industry benchmarks sourced
  □ Peer comparisons available
  □ Best-in-class targets defined

□ Actionability
  □ Key insights documented
  □ Root causes investigated
  □ Recommendations developed
  □ Actions assigned and tracked

Anti-Patterns

• Single cohort obsession — Focusing on one segment without context
• Insufficient sample size — Drawing conclusions from tiny cohorts
• Ignoring seasonality — Comparing Q4 to Q1 without adjustment
• Inconsistent definitions — Changing metrics mid-analysis
• Survivorship bias — Only analyzing retained customers
• No benchmarks — Can't assess "good" without comparison
• Analysis paralysis — Too many cohorts, no action
• Stale analysis — Running cohort analysis once, never updating

Customer Data Enrichment & 360 View

Impact: HIGH

A complete customer view is the foundation of effective health scoring and risk prediction. Data enrichment fills gaps, adds context, and creates a unified picture that enables proactive customer success. Without comprehensive data, even the best health models fail.

The Customer 360 Architecture

┌──────────────────────────────────────────────────────────────────┐
│                      CUSTOMER 360 VIEW                           │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐              │
│  │   COMPANY   │  │  CONTACTS   │  │  CONTRACT   │              │
│  │   PROFILE   │  │  & ROLES    │  │   DETAILS   │              │
│  └─────────────┘  └─────────────┘  └─────────────┘              │
│                                                                  │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐              │
│  │   PRODUCT   │  │   SUPPORT   │  │   BILLING   │              │
│  │   USAGE     │  │   HISTORY   │  │   HISTORY   │              │
│  └─────────────┘  └─────────────┘  └─────────────┘              │
│                                                                  │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐              │
│  │   COMMS     │  │   SUCCESS   │  │  EXTERNAL   │              │
│  │   HISTORY   │  │   METRICS   │  │   SIGNALS   │              │
│  └─────────────┘  └─────────────┘  └─────────────┘              │
│                                                                  │
│                    ┌─────────────────┐                          │
│                    │   HEALTH SCORE  │                          │
│                    │   & RISK MODEL  │                          │
│                    └─────────────────┘                          │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Data Source Categories

Good Data Enrichment Strategy

Customer Profile: Acme Corp

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

COMPANY INFORMATION (Enriched)
├── Legal Name: Acme Corporation
├── Industry: SaaS / B2B Technology
├── Employee Count: 450 (↑ 15% YoY)
├── Annual Revenue: $45M (estimated)
├── Funding: Series C, $28M raised
├── HQ Location: San Francisco, CA
├── Founded: 2018
└── Growth Stage: Scale-up

TECHNOGRAPHICS
├── CRM: Salesforce
├── Marketing: HubSpot
├── Support: Zendesk
├── Analytics: Mixpanel
├── Integrations Active: Salesforce, Slack
└── Potential Integrations: HubSpot, Zendesk

CONTACT INTELLIGENCE
├── Decision Makers: 3 identified
├── Champion: Sarah Chen (Head of Ops)
├── Executive Sponsor: Michael Torres (VP)
├── Billing Contact: Finance team
├── Power Users: 8 identified
└── Stakeholder Health: Strong

INTENT SIGNALS
├── Competitor Research: None detected
├── Content Engagement: 12 articles last month
├── Webinar Attendance: Attended 2 of 3 offered
└── Community Activity: Active in user group

EXTERNAL SIGNALS
├── Recent News: Announced new product line
├── Job Postings: Hiring 3 ops roles (expansion signal)
├── LinkedIn Activity: Champion posted about our product
└── Sentiment: Positive social mentions

DERIVED INSIGHTS
├── Expansion Potential: High (hiring, growing)
├── Churn Risk Factors: None detected
├── Recommended Actions: Upsell conversation
└── Next Best Action: Schedule expansion QBR

Bad Data Enrichment Strategy

Customer Profile: Acme Corp

Company Name: Acme Corp
Contact: Sarah
Email: sarah@acme.com
Plan: Enterprise
MRR: $10,000

Problems:
✗ Minimal company context
✗ No firmographic enrichment
✗ No contact role or hierarchy
✗ No intent or external signals
✗ No usage data integration
✗ No derived insights
✗ No next best action
✗ Static, not dynamic data

Key Enrichment Fields

Contact Enrichment Strategy

Contact Hierarchy Mapping:

Executive Level
├── CEO: John Smith
│   └── Relationship: Met once, annual review
├── CFO: Lisa Wong
│   └── Relationship: Billing escalations only
└── VP Operations: Michael Torres (Exec Sponsor)
    └── Relationship: Monthly check-ins ✓

Management Level
├── Head of Ops: Sarah Chen (Champion)
│   └── Relationship: Weekly calls ✓
├── IT Director: David Park
│   └── Relationship: Technical contact
└── Finance Manager: Amy Liu
    └── Relationship: Billing contact

User Level
├── Power Users: 8 identified
├── Regular Users: 23 active
└── Dormant Users: 4 inactive

Stakeholder Health Score: 78/100
├── Champion strength: Strong
├── Multi-threading: Good (4 relationships)
├── Executive access: Moderate
└── Risk: Champion single point of failure

Data Quality Framework

Data Quality Dashboard

Customer Data Quality Report

┌─────────────────────────────────────────────────────────────────┐
│  OVERALL DATA QUALITY SCORE: 81%                                │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  Completeness by Category:                                      │
│  ├── Company Profile:     ████████████████████ 92%             │
│  ├── Contact Data:        █████████████████░░░ 78%             │
│  ├── Product Usage:       ████████████████████ 96%             │
│  ├── Support History:     █████████████████░░░ 82%             │
│  ├── External Signals:    ████████████░░░░░░░░ 58%             │
│  └── Financial Data:      ██████████████████░░ 89%             │
│                                                                 │
│  Data Freshness:                                                │
│  ├── Updated <7 days:     65% of accounts                      │
│  ├── Updated 8-30 days:   28% of accounts                      │
│  └── Updated >30 days:    7% of accounts (⚠ stale)            │
│                                                                 │
│  Data Issues:                                                   │
│  ├── Missing champion:    23 accounts                          │
│  ├── Invalid email:       12 contacts                          │
│  ├── Duplicate contacts:  8 records                            │
│  └── Stale firmographics: 34 accounts                          │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Integration Architecture

Data Governance Principles

1. Single Source of Truth
   - Define master system for each data type
   - Health score is calculated, not stored in CRM
   - CRM is master for relationships
   - Product database is master for usage

2. Ownership
   - Each data field has a defined owner
   - Owner responsible for quality
   - Regular audits by data team

3. Access Control
   - Sensitive data (PII) protected
   - Role-based access
   - Audit logging enabled

4. Privacy Compliance
   - GDPR / CCPA compliant enrichment
   - Consent management
   - Data retention policies
   - Right to deletion supported

Enrichment ROI Calculation

Data Enrichment Checklist

□ Core Customer Data
  □ Company name and legal entity
  □ Industry and sub-industry
  □ Employee count and trend
  □ Location (HQ and offices)
  □ Website and social profiles

□ Contact Data
  □ Key contacts identified
  □ Roles and hierarchy mapped
  □ Email and phone validated
  □ LinkedIn profiles linked
  □ Champion and sponsor flagged

□ Financial Data
  □ Contract details accurate
  □ MRR/ARR calculated correctly
  □ Payment history current
  □ Renewal dates tracked
  □ Expansion history captured

□ Behavioral Data
  □ Product usage integrated
  □ Support tickets linked
  □ Communication history captured
  □ Engagement metrics calculated
  □ Feature adoption tracked

□ External Signals
  □ Firmographic enrichment active
  □ Intent data flowing
  □ News monitoring enabled
  □ Job posting tracking
  □ Social sentiment captured

□ Data Quality
  □ Completeness monitored
  □ Freshness tracked
  □ Duplicates resolved
  □ Validation rules in place
  □ Regular audits scheduled

Anti-Patterns

• Data silos — Product data separate from CRM separate from support
• Manual enrichment — Relying on CSMs to research and update
• Stale data — Firmographics from years ago
• Over-collection — Gathering data without clear use case
• No single source of truth — Conflicting data across systems
• Privacy violations — Enriching without consent
• Ignoring data quality — Garbage in, garbage out
• Under-utilization — Rich data not surfaced to users

Executive Reporting & Dashboards

Impact: HIGH

Executive reporting transforms customer health data into strategic business insights. The goal isn't just presenting metrics — it's enabling better decisions about customer investments, product direction, and company strategy. The best reports tell a story that drives action.

The Executive Reporting Hierarchy

┌──────────────────────────────────────────────────────────────────┐
│                  REPORTING HIERARCHY                             │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│  BOARD LEVEL                                                     │
│  └── High-level health, NRR, strategic risks                    │
│      Frequency: Quarterly                                        │
│                                                                  │
│  C-SUITE LEVEL                                                   │
│  └── Portfolio health, trends, strategic accounts               │
│      Frequency: Monthly                                          │
│                                                                  │
│  VP/DIRECTOR LEVEL                                               │
│  └── Team performance, segment health, initiatives              │
│      Frequency: Weekly                                           │
│                                                                  │
│  MANAGER LEVEL                                                   │
│  └── Individual accounts, risk alerts, action items             │
│      Frequency: Daily                                            │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Key Executive KPIs

Good Executive Dashboard

Customer Success Executive Dashboard
Period: January 2025

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

PORTFOLIO HEALTH SUMMARY

Total ARR:           $24.5M      Health Score Avg:    72 (↑ 3)
Customers:           485         At-Risk ARR:         $2.1M (8.6%)
NRR (Trailing 12M):  112%        Time to Value:       22 days

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

REVENUE METRICS                           vs. Prior Month

Gross Retention:     92%                  ↑ +1%
Net Retention:       108%                 ↑ +2%
Expansion Revenue:   $412K                ↑ +15%
Churned Revenue:     $198K                ↓ -22% (improvement)
Contraction:         $89K                 ↓ -8% (improvement)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

HEALTH DISTRIBUTION

Thriving (85+):   ████████████░░░░░░░░  28% ($6.9M)
Healthy (70-84):  ██████████████░░░░░░  38% ($9.3M)
Neutral (50-69):  ████████░░░░░░░░░░░░  22% ($5.4M)
At-Risk (30-49):  ████░░░░░░░░░░░░░░░░  9%  ($2.2M)
Critical (<30):   █░░░░░░░░░░░░░░░░░░░  3%  ($0.7M)

Trend: Distribution improving (at-risk down from 12%)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

STRATEGIC ACCOUNTS STATUS (Top 20 by ARR)

Green:  14  accounts ($8.2M ARR)
Yellow: 4   accounts ($2.1M ARR)
Red:    2   accounts ($1.4M ARR)  ← Executive attention required

Red Accounts:
1. GlobalTech Inc ($850K) - Champion departure, exec engaged
2. MegaCorp ($550K) - Competitive threat, QBR scheduled

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

KEY WINS THIS MONTH
✓ Saved $420K at-risk ARR (TechFlow, DataPro)
✓ Closed $380K expansion (Acme Corp +$150K, 3 others)
✓ NPS improved 8 points (32 → 40)

KEY RISKS TO WATCH
⚠ 3 renewals >$100K in next 60 days at health <60
⚠ Enterprise segment NPS declined 5 points
⚠ Q2 cohort showing early retention weakness

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Bad Executive Dashboard

Customer Success Report - January

Customers: 485
ARR: $24,500,000
Health Score: 72
NPS: 40

Churned: 12 customers
New: 28 customers

Support Tickets: 1,247

Problems:
✗ No context or trends
✗ No targets or benchmarks
✗ No segmentation
✗ No actionable insights
✗ Mixing operational and strategic metrics
✗ No risk visibility
✗ No narrative
✗ No recommendations

Board-Level Reporting

Board Report: Customer Success (Q4 2024)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

EXECUTIVE SUMMARY

Net Revenue Retention of 112% demonstrates strong customer health
and expansion motion. At-risk ARR has decreased 25% since Q3,
indicating improved early intervention effectiveness.

Key achievements:
• Reduced churn rate from 1.8% to 1.2% monthly
• Expanded NRR from 105% to 112%
• Decreased time-to-value from 34 to 22 days

Areas requiring investment:
• Enterprise segment engagement (NPS declining)
• Proactive risk detection (surprise churn rate 18%)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

KEY METRICS

                    Q4 2024    Q3 2024    YoY       Target
Net Revenue Ret.    112%       105%       +18%      110%  ✓
Gross Revenue Ret.  92%        90%        +4%       90%   ✓
Logo Retention      94%        93%        +2%       92%   ✓
NPS                 40         32         +12       35    ✓

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

STRATEGIC RISKS

1. Enterprise Engagement (Medium Risk)
   - NPS declined 5 points in segment
   - Two $500K+ accounts in yellow status
   - Mitigation: Executive business reviews, product investment

2. Market Competition (Low-Medium Risk)
   - Competitor mentions up 15% in support tickets
   - No significant losses yet
   - Mitigation: Competitive intelligence program launched

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Q1 2025 PRIORITIES

1. Launch enterprise engagement program
2. Reduce surprise churn rate to <10%
3. Achieve 115% NRR target

Dashboard Design Principles

Data Storytelling Framework

The Situation-Complication-Resolution Framework

SITUATION (What's happening)
"Our customer portfolio grew 22% this year to $24.5M ARR
across 485 customers."

COMPLICATION (Why it matters)
"However, our at-risk ARR has increased to $2.1M (8.6%),
driven primarily by declining engagement in the enterprise
segment where NPS dropped 5 points."

RESOLUTION (What we're doing)
"We're launching a dedicated enterprise success program
with executive business reviews, which has shown 40%
improvement in similar situations. Expected impact:
reduce at-risk enterprise ARR by 50% in Q1."

KEY INSIGHT
Lead with the insight, not the data.

Bad: "Health scores averaged 72 this month."
Good: "Customer health improved for the 3rd consecutive month,
      driven by our new onboarding program which reduced
      time-to-value by 35%."

Reporting Cadence

Dashboard Metrics by Audience

Report Automation

Executive Presentation Checklist

□ Pre-Meeting Preparation
  □ Data refreshed and validated
  □ Key metrics calculated correctly
  □ Narrative prepared and reviewed
  □ Anticipated questions researched
  □ Backup slides ready

□ Content Structure
  □ Executive summary on first slide
  □ Key metrics with context
  □ Trends and comparisons shown
  □ Strategic risks highlighted
  □ Wins and successes celebrated
  □ Clear recommendations included
  □ Ask/investment needs specified

□ Visual Design
  □ Consistent formatting
  □ Clear hierarchy
  □ Minimal clutter
  □ Actionable insights highlighted
  □ Red/yellow/green status clear

□ Delivery
  □ Lead with insights, not data
  □ Tell a story
  □ Acknowledge challenges honestly
  □ Provide recommendations
  □ Allow time for questions
  □ Document action items

Good vs Bad Metrics Presentation

Anti-Patterns

• Data dump — Too many metrics without narrative
• No benchmarks — Metrics without targets or comparisons
• Vanity focus — Highlighting good metrics, hiding problems
• Stale reporting — Manual processes creating delays
• One-size-fits-all — Same report for board and manager
• No action items — Reporting without recommendations
• Surprise reveals — Board learns about risks first in meeting
• Metric overload — 50 KPIs when 5 would suffice

Health Model Validation & Calibration

Impact: HIGH

A health score model is only valuable if it accurately predicts outcomes. Without regular validation and calibration, models drift, accuracy degrades, and teams lose confidence. Continuous validation ensures your health scores remain actionable and trustworthy.

The Validation Lifecycle

┌──────────────────────────────────────────────────────────────────┐
│                  MODEL VALIDATION LIFECYCLE                      │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│       ┌─────────┐                                                │
│       │  BUILD  │                                                │
│       └────┬────┘                                                │
│            │                                                     │
│            ▼                                                     │
│       ┌─────────┐      ┌─────────┐      ┌─────────┐            │
│       │ DEPLOY  │ ───► │ MONITOR │ ───► │ ANALYZE │            │
│       └─────────┘      └────┬────┘      └────┬────┘            │
│            ▲                │                │                  │
│            │                ▼                ▼                  │
│            │           ┌─────────┐      ┌─────────┐            │
│            └───────────│ REFINE  │ ◄─── │ CALIBRATE│            │
│                        └─────────┘      └─────────┘            │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Key Validation Metrics

Good Model Validation Report

Health Score Model Validation Report
Period: Q4 2024

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

EXECUTIVE SUMMARY

Model performance meets targets across key metrics.
Prediction accuracy improved 8% vs. Q3 following
feature updates. One area of concern: enterprise
segment showing higher surprise churn rate.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

PREDICTION ACCURACY

Actual Churns:               47
Predicted (Health <50):      38
Correctly Predicted:         33
Surprise Churns (>60):       14

Accuracy Rate:               70% (target: 70%) ✓
Surprise Churn Rate:         30% (target: <20%) ⚠
False Positive Rate:         28% (target: <30%) ✓

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SCORE-OUTCOME CORRELATION

                    Retention    Expansion    NPS
Health Score        0.62         0.48         0.55
Correlation         Strong       Moderate     Moderate

                    Prior Quarter   Current
Retention Corr.     0.58            0.62  ↑
Expansion Corr.     0.45            0.48  ↑
NPS Corr.           0.52            0.55  ↑

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

LIFT ANALYSIS

Decile    Avg Score    Churn Rate    Lift
1 (High)  12           42%           5.2x  ← Good separation
2         28           31%           3.9x
3         38           22%           2.8x
4         48           15%           1.9x
5         56           11%           1.4x
6         63           9%            1.1x
7         70           7%            0.9x
8         76           5%            0.6x
9         83           3%            0.4x
10 (Low)  91           1%            0.1x

Overall churn rate: 8%
Top decile lift: 5.2x (target: >3x) ✓

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SEGMENT ANALYSIS

Segment         Accuracy    Surprise Rate    Status
Enterprise      62%         38%              ⚠ Needs attention
Mid-Market      74%         22%              ✓ On track
SMB             72%         28%              ✓ On track
Startup         68%         32%              ~ Monitor

Enterprise segment investigation:
- 5 of 14 surprise churns were enterprise
- Common pattern: Champion departure not detected
- Recommendation: Add LinkedIn monitoring signal

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

CALIBRATION CHECK

Score Range    Predicted Risk    Actual Risk    Gap
0-20           85%               78%            -7%
20-40          60%               55%            -5%
40-60          35%               32%            -3%
60-80          15%               12%            -3%
80-100         5%                4%             -1%

Avg Calibration Error: 4% (target: <5%) ✓
Model is slightly overconfident in high-risk scores.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

RECOMMENDATIONS

1. [HIGH] Add champion monitoring to enterprise scoring
2. [MEDIUM] Recalibrate high-risk thresholds
3. [LOW] Review startup segment feature weights

Bad Model Validation Report

Health Score Report

Model accuracy: 70%
Churns predicted: 33/47

Status: Working fine.

Problems:
✗ No trend analysis
✗ No segment breakdown
✗ No calibration check
✗ No feature analysis
✗ No actionable recommendations
✗ No comparison to prior period
✗ No investigation of failures

Surprise Churn Analysis Framework

Surprise Churn Investigation: AccountName

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

ACCOUNT PROFILE
├── ARR: $85,000
├── Tenure: 18 months
├── Health Score at Churn: 72 (Healthy)
├── Health Score 30 days prior: 74
└── Health Score 90 days prior: 71

CHURN REASON: Competitor displacement

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SIGNAL ANALYSIS: WHAT WE MISSED

Signal                  Present?    In Model?    Why Missed?
─────────────────────────────────────────────────────────────
Competitor research     Yes         No           No intent data
Champion job search     Yes         No           No LinkedIn tracking
Reduced engagement      Subtle      Yes          Below threshold
Support complaints      No          -            No signal
Usage decline           Minor       Yes          Below threshold

ROOT CAUSE: Champion was evaluating alternatives
while maintaining appearance of engagement.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

MODEL IMPROVEMENT OPPORTUNITIES

1. Add intent data signals (competitor research)
2. Add LinkedIn monitoring for key contacts
3. Lower threshold for engagement decline
4. Create composite "quiet leaving" indicator
5. Weight recent trend more heavily

EXPECTED IMPACT: Could have caught this 60 days earlier

Calibration Techniques

Model Drift Detection

Drift Monitoring Dashboard

┌─────────────────────────────────────────────────────────────────┐
│  FEATURE DRIFT MONITORING                                       │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  Feature                   Baseline    Current     Drift       │
│  ────────────────────────────────────────────────────────────  │
│  Usage velocity (30d)      -0.02       -0.08       ⚠ DRIFT    │
│  NPS score                 42          38          ~ Minor     │
│  Support tickets/mo        2.3         2.5         ✓ Stable   │
│  Feature adoption          58%         55%         ✓ Stable   │
│  Champion engagement       0.72        0.68        ~ Minor     │
│                                                                 │
│  ⚠ Alert: Usage velocity distribution shifted significantly   │
│     Recommend: Investigate cause, consider retraining          │
│                                                                 │
├─────────────────────────────────────────────────────────────────┤
│  OUTCOME DRIFT                                                  │
│                                                                 │
│  Metric                    Baseline    Current     Status      │
│  ────────────────────────────────────────────────────────────  │
│  Monthly churn rate        1.5%        1.8%        ~ Monitor   │
│  Score-churn correlation   0.62        0.58        ~ Monitor   │
│  Prediction accuracy       72%         68%         ⚠ Watch    │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Validation Schedule

Threshold Calibration Process

Step 1: Analyze current distribution
├── Plot health scores vs. outcomes
├── Identify natural breakpoints
└── Calculate churn rate by score band

Step 2: Assess operational capacity
├── How many at-risk accounts can CSMs handle?
├── What's the cost of false positives?
└── What's the cost of missed churns?

Step 3: Optimize thresholds
├── Set thresholds to balance precision/recall
├── Consider segment-specific adjustments
└── Align with intervention capacity

Step 4: Validate proposed changes
├── Backtest on historical data
├── Calculate expected false positive/negative rates
└── Estimate resource requirements

Step 5: Implement and monitor
├── Update threshold configuration
├── Communicate to CS team
├── Track performance post-change
└── Adjust if needed

Feature Importance Review

Model Governance Checklist

□ Validation Process
  □ Weekly accuracy tracking automated
  □ Surprise churn review process defined
  □ Drift alerts configured
  □ Quarterly full validation scheduled

□ Documentation
  □ Model architecture documented
  □ Feature definitions captured
  □ Threshold rationale recorded
  □ Version history maintained

□ Change Management
  □ Change approval process defined
  □ A/B testing capability available
  □ Rollback plan documented
  □ Communication plan for changes

□ Stakeholder Alignment
  □ CS leadership reviews validation reports
  □ Data team owns model maintenance
  □ Feedback loop from CSMs formalized
  □ Executive sponsor engaged

□ Continuous Improvement
  □ New feature experimentation process
  □ Segment-specific tuning allowed
  □ Industry benchmark tracking
  □ Model improvement backlog maintained

Anti-Patterns

• Set and forget — Never validating after initial launch
• Aggregate-only analysis — Missing segment-specific issues
• No surprise churn investigation — Not learning from failures
• Threshold stagnation — Never adjusting as business changes
• Ignoring drift — Features change meaning over time
• No documentation — Model logic in one person's head
• Validation without action — Reports with no follow-through
• Perfect-seeking — Waiting for 100% accuracy vs. iterating

Health Score Design & Weighting

Impact: CRITICAL

A well-designed health score predicts customer outcomes 60-90 days before they happen. Poor health scores are vanity metrics that provide false confidence while customers silently churn.

The Health Score Equation

Health Score = Σ (Component Score × Weight)

Where:
- Each component is normalized to 0-100
- Weights sum to 100%
- Final score ranges 0-100

Component Selection Criteria

Standard Health Score Components

Weight Assignment by Business Model

Good Health Score Design

Health Score v2.0 - Enterprise Accounts

Component: Product Usage (35%)
├── DAU/MAU ratio (10%)
│   └── 30-day rolling average
├── Feature adoption score (10%)
│   └── % of key features used
├── Usage depth (10%)
│   └── Actions per session
└── Core workflow completion (5%)
    └── % completing primary use case

Component: Engagement (25%)
├── Relationship NPS (10%)
│   └── Most recent score
├── CSM touchpoints (8%)
│   └── Meetings held vs. scheduled
└── Communication responsiveness (7%)
    └── Email response rate

Component: Growth Signals (20%)
├── Seat expansion trend (8%)
│   └── 90-day user growth rate
├── Usage expansion trend (7%)
│   └── 90-day consumption growth
└── Contract expansion (5%)
    └── Any expansion in last year

Component: Support Health (20%)
├── Ticket sentiment (8%)
│   └── AI-analyzed support conversations
├── Resolution satisfaction (7%)
│   └── Post-ticket CSAT
└── Escalation frequency (5%)
    └── Escalations per month

Scoring:
- All sub-metrics normalized to 0-100
- Component score = weighted average of sub-metrics
- Final score = weighted sum of components

Bad Health Score Design

Health Score v1.0 (Problems Identified)

Components:
├── Product Usage (70%)          ← Over-weighted single category
│   └── Total logins             ← Vanity metric
│
├── Support Tickets (15%)        ← Direction unclear
│   └── Total tickets opened     ← More tickets = lower score?
│
└── Contract Value (15%)         ← Not predictive
    └── ARR                      ← Bigger customers ≠ healthier

Problems:
✗ Over-reliance on single category
✗ Logins don't measure value
✗ Tickets could be good (engaged) or bad (frustrated)
✗ ARR doesn't predict retention
✗ No engagement or relationship signals
✗ No leading indicators

Scoring Algorithm Examples

Linear Scoring:

Score = (Actual Value / Target Value) × 100
Cap at 100, floor at 0

Example: DAU/MAU
Target: 40%
Actual: 32%
Score: (32/40) × 100 = 80

Threshold-Based Scoring:

If DAU/MAU >= 50%: Score = 100
If DAU/MAU >= 40%: Score = 80
If DAU/MAU >= 30%: Score = 60
If DAU/MAU >= 20%: Score = 40
If DAU/MAU >= 10%: Score = 20
If DAU/MAU < 10%:  Score = 0

Trend-Adjusted Scoring:

Base Score = Current metric score
Trend Factor = (Current - 30 days ago) / 30 days ago
Adjusted Score = Base Score × (1 + Trend Factor × 0.2)

Example:
Base Score: 70
Usage up 15%: 70 × 1.03 = 72.1
Usage down 15%: 70 × 0.97 = 67.9

Health Score Thresholds

Threshold Calibration Process

Step 1: Historical Analysis
- Pull 12+ months of health scores
- Tag customers by outcome (churned, retained, expanded)
- Plot score distribution by outcome

Step 2: Threshold Identification
- Find score ranges where outcomes diverge
- Identify clear "danger zones"
- Map to intervention capacity

Step 3: Validation
- Apply thresholds prospectively
- Track prediction accuracy
- Measure false positive/negative rates

Step 4: Refinement
- Adjust thresholds quarterly
- Segment-specific thresholds if needed
- Document rationale for changes

Health Score Validation Metrics

Segment-Specific Scoring Considerations

Health Score Implementation Checklist

□ Component Selection
  □ Each component has clear predictive value
  □ All data sources are reliable and available
  □ Metrics are actionable (we can influence them)
  □ No duplicate signals across components

□ Weight Assignment
  □ Weights based on historical correlation analysis
  □ Weights sum to 100%
  □ No single component dominates (max 40%)
  □ Weights documented with rationale

□ Scoring Logic
  □ All sub-metrics normalized consistently (0-100)
  □ Handling for missing data defined
  □ Edge cases documented (new customers, etc.)
  □ Calculation logic peer-reviewed

□ Threshold Definition
  □ Thresholds based on outcome analysis
  □ Clear actions mapped to each threshold
  □ Thresholds validated against historical data
  □ Segment-specific adjustments if needed

□ Operational Readiness
  □ Score calculation automated
  □ Update frequency defined (daily/weekly)
  □ Alerting configured for threshold crossings
  □ Dashboard visibility for CS team

□ Ongoing Governance
  □ Quarterly calibration review scheduled
  □ Accuracy metrics tracked
  □ Feedback loop from CS team
  □ Version history maintained

Anti-Patterns

• Kitchen sink scoring — Including every metric regardless of predictive value
• Equal weighting — All components at 20% without analysis
• Binary signals — Using yes/no when degree matters
• Static thresholds — Never recalibrating as business changes
• Ignoring tenure — New customers scored same as mature ones
• Vanity components — Metrics that feel important but don't predict
• Over-fitting — Optimizing for historical data, failing on new patterns
• No documentation — Scoring logic understood by one person only

Leading vs Lagging Indicator Analysis

Impact: CRITICAL

By the time you see lagging indicators (churn, downgrades), it's often too late. Leading indicators give you the 60-90 day window needed to intervene effectively. The best customer success teams obsess over leading indicators.

Indicator Classification Framework

Timeline to Outcome:
────────────────────────────────────────────────────────►
│                                                       │
│  LEADING           COINCIDENT         LAGGING        │
│  (60-90 days)      (30-60 days)       (0-30 days)   │
│                                                       │
│  ✓ Actionable      ~ Urgent           ✗ Historical  │
│  ✓ Predictive      ~ Confirmatory     ✗ Reactive    │
│  ✓ Proactive       ~ Responsive       ✗ Post-mortem │
│                                                       │
└───────────────────────────────────────────────────────┘

Common Indicator Categories

Leading Indicator Catalog

Good Leading Indicator Analysis

Indicator: Feature Adoption Decline

Definition:
- Customer using <50% of features used at peak
- Measured over rolling 30-day window
- Compared to their own historical baseline

Why It's Leading:
- Precedes churn by 75 days on average
- Indicates value not being realized
- Actionable through enablement

Detection:
- Automated daily feature usage calculation
- Alert when adoption drops below threshold
- Trend visualization in health dashboard

Correlation Analysis:
- 68% of customers with this signal churned within 120 days
- Only 12% of customers without this signal churned
- Predictive accuracy: 73%

Action Trigger:
When detected → CSM outreach within 48 hours
Goal → Feature re-enablement or use case pivot

Bad Leading Indicator Analysis

Indicator: Low NPS Score

Problems:
✗ NPS is often coincident or lagging, not leading
✗ By the time NPS drops, issues are entrenched
✗ Quarterly surveys miss the window
✗ NPS alone lacks actionability

Better Approach:
- Track NPS trend (leading signal: declining NPS)
- Combine with other signals (NPS + usage decline)
- Use transactional NPS for faster feedback
- Look at verbatim comments for leading signals

Correlation Reality:
- Static low NPS: 45% correlation to churn
- Declining NPS trend: 72% correlation to churn
- The trend is the leading indicator, not the score

Correlation Analysis Methodology

Step 1: Define Outcomes
- Primary: Churn (Y/N)
- Secondary: Expansion, Contraction, NRR

Step 2: Identify Candidate Signals
- List all measurable customer behaviors
- Include product, engagement, support, financial

Step 3: Time-Shift Analysis
For each signal at each lag period (30, 60, 90, 120 days):
- Calculate correlation to outcome
- Identify optimal prediction window

Step 4: Signal Ranking
- Rank by correlation strength
- Consider actionability
- Assess data availability

Step 5: Combine for Prediction
- Build composite leading indicator score
- Validate on holdout data
- Monitor ongoing accuracy

Correlation Strength Benchmarks

Signal Combination Matrix

Good Indicator Monitoring Dashboard

Leading Indicator Dashboard

┌─────────────────────────────────────────────────────────┐
│  LEADING INDICATOR ALERTS (Last 7 Days)                 │
├─────────────────────────────────────────────────────────┤
│                                                         │
│  Critical (Action Required):                   12       │
│  ├── Usage Decline >30%                        5       │
│  ├── Champion Departure Detected               3       │
│  └── Renewal Meeting Not Scheduled             4       │
│                                                         │
│  Warning (Monitor Closely):                    28       │
│  ├── Feature Adoption Declining                12      │
│  ├── Engagement Score Down                     9       │
│  └── Support Sentiment Shift                   7       │
│                                                         │
├─────────────────────────────────────────────────────────┤
│  INDICATOR TRENDS (90-Day)                              │
│                                                         │
│  Usage Decline Alerts:     ▲ +15% vs prior period      │
│  Champion Departures:      ▼ -8% vs prior period       │
│  Engagement Drops:         ─ Flat vs prior period      │
│                                                         │
├─────────────────────────────────────────────────────────┤
│  PREDICTION ACCURACY (Last Quarter)                     │
│                                                         │
│  Churns Predicted:         42/51 (82%)                 │
│  False Positives:          15/42 (36%)                 │
│  Avg Lead Time:            67 days                     │
│                                                         │
└─────────────────────────────────────────────────────────┘

Building Your Leading Indicator Model

Action Triggers by Signal

Indicator Validation Checklist

□ Predictive Power
  □ Correlation to outcome >0.5
  □ Consistent across customer segments
  □ Maintains accuracy over time
  □ Not just correlating with another signal

□ Actionability
  □ Clear intervention available
  □ Enough lead time to act (60+ days)
  □ Team has capacity to respond
  □ Success interventions documented

□ Reliability
  □ Data source is consistent
  □ Signal can be calculated automatically
  □ Missing data handling defined
  □ False positive rate acceptable (<30%)

□ Operationalization
  □ Real-time or near-real-time detection
  □ Alerts configured and routed correctly
  □ Playbook exists for each signal
  □ Feedback loop to improve model

Anti-Patterns

• Lagging indicator focus — Tracking churn rate instead of churn predictors
• Single indicator reliance — One signal without confirmation
• Ignoring signal combinations — Missing that A + B together is critical
• Static thresholds — Not adjusting for segment or seasonality
• No validation — Using indicators without testing predictive power
• Action-less alerts — Signals without defined responses
• Too many indicators — Alert fatigue from over-monitoring
• Ignoring false positives — Not refining to reduce noise

Risk Identification & Escalation

Impact: CRITICAL

Early risk identification and well-defined escalation processes are the difference between saving an at-risk account and conducting a post-mortem. A structured approach ensures no customer falls through the cracks and interventions happen with enough lead time to succeed.

The Risk Escalation Framework

┌──────────────────────────────────────────────────────────────────┐
│                     RISK ESCALATION PATH                         │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│  LOW RISK          MEDIUM RISK        HIGH RISK        CRITICAL │
│  Health: 70+       Health: 50-69      Health: 30-49    Health: <30│
│                                                                  │
│  ┌─────────┐      ┌─────────┐       ┌─────────┐      ┌─────────┐│
│  │ Monitor │ ───► │ Engage  │ ────► │Intervene│ ───► │Escalate ││
│  └─────────┘      └─────────┘       └─────────┘      └─────────┘│
│                                                                  │
│  Owner: CSM       Owner: CSM        Owner: CSM +     Owner: VP + │
│                                     Manager          Executive   │
│                                                                  │
│  SLA: Weekly      SLA: 1 week       SLA: 48 hours   SLA: 24 hrs │
│  review           outreach          intervention    response     │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Risk Signal Categories

Risk Signal Severity Matrix

Good Risk Identification System

Risk Alert: Acme Corp

Account: Acme Corp
ARR: $125,000
Health Score: 42 (was 68 last month)
CSM: Jane Smith

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

RISK SIGNALS DETECTED:

1. Champion Status Change (CRITICAL)
   └── Sarah Johnson updated LinkedIn to new company
   └── Detected: 2 hours ago
   └── She represented 65% of account activity

2. Usage Decline (HIGH)
   └── 34% decrease in DAU over 30 days
   └── Key feature "Reports" unused for 14 days
   └── Trend accelerating

3. Support Sentiment (MEDIUM)
   └── Last 3 tickets rated "Dissatisfied"
   └── Average sentiment score: 2.1/5 (was 4.2)

RISK SCORE: 78/100 (High Risk)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

RECOMMENDED ACTIONS:

1. [IMMEDIATE] Contact account to identify new champion
2. [48 HOURS] Schedule executive check-in
3. [1 WEEK] Arrange re-onboarding for new stakeholders

ESCALATION: Manager + VP CS notified
SLA: Response required within 24 hours

Bad Risk Identification System

Alert: Account health decreased

Account: Acme Corp
Health Score: 42
Alert: Health score below threshold

Problems:
✗ No specific signals identified
✗ No context on what changed
✗ No severity classification
✗ No recommended actions
✗ No escalation path
✗ No SLA defined
✗ No ARR/impact context

Escalation Matrix

Intervention Playbooks

Playbook: Champion Departure

Trigger: Key contact leaves company
Severity: Critical
SLA: 24 hour initial response

Day 1:
□ Verify departure (LinkedIn, email bounce, etc.)
□ Identify replacement contact
□ Executive-to-executive outreach to maintain relationship
□ Update CRM with new stakeholder map

Day 2-7:
□ Schedule intro call with new champion
□ Offer re-onboarding/training
□ QBR to re-establish value baseline
□ Document new success criteria

Day 8-30:
□ Accelerate engagement cadence
□ Monthly check-ins (vs. quarterly)
□ Feature adoption review
□ Executive sponsor assignment if needed

Playbook: Usage Decline

Trigger: Usage down >25% over 30 days
Severity: High
SLA: 48 hour initial contact

Day 1-2:
□ Analyze usage data for root cause
□ Identify which users/features affected
□ CSM outreach: "I noticed [specific change], is everything okay?"
□ Offer support call

Day 3-7:
□ Deep-dive call to understand context
□ Create action plan with customer
□ Enablement session if adoption issue
□ Executive involvement if strategic issue

Day 8-30:
□ Weekly check-ins during recovery
□ Monitor usage daily
□ Adjust plan based on progress
□ Escalate if no improvement by day 14

Playbook: Competitive Threat

Trigger: Competitor mention detected
Severity: Very High
SLA: 48 hour executive response

Day 1:
□ Alert CSM, Manager, and VP
□ Gather intelligence (what competitor, why looking)
□ Prepare competitive battle card
□ Schedule executive call

Day 2-3:
□ Executive-to-executive engagement
□ Understand specific evaluation criteria
□ Address gaps or concerns directly
□ Reinforce unique value proposition

Day 4-14:
□ Provide additional proof points (case studies, ROI)
□ Offer executive references
□ Consider strategic concessions if needed
□ Document outcome and learnings

Save Team Structure

Save Offer Guidelines

Risk Review Cadence

Risk Documentation Template

## At-Risk Account Analysis

**Account:** [Name]
**ARR:** [Amount]
**Health Score:** [Current] (was [Previous])
**Risk Level:** [Critical/High/Medium]
**Date Identified:** [Date]

### Risk Signals
| Signal | Severity | Date Detected |
|--------|----------|---------------|
| [Signal 1] | [Level] | [Date] |
| [Signal 2] | [Level] | [Date] |

### Root Cause Analysis
[What's driving the risk]

### Stakeholder Impact
- Champion: [Status]
- Executive Sponsor: [Status]
- End Users: [Status]

### Action Plan
| Action | Owner | Due Date | Status |
|--------|-------|----------|--------|
| [Action 1] | [Name] | [Date] | [Status] |
| [Action 2] | [Name] | [Date] | [Status] |

### Outcome
[ ] Saved
[ ] Churned
[ ] In Progress

### Lessons Learned
[What we'll do differently]

Escalation Checklist

□ Risk Identification
  □ Specific signals documented
  □ Severity classified correctly
  □ Root cause hypothesized
  □ ARR impact quantified

□ Initial Response
  □ CSM contacted within SLA
  □ Customer context gathered
  □ Quick win opportunities identified
  □ Escalation need assessed

□ Escalation Execution
  □ Right people involved
  □ Clear ask defined
  □ Timeline established
  □ Customer expectations set

□ Intervention
  □ Action plan created
  □ Customer agreement obtained
  □ Progress tracking in place
  □ Success criteria defined

□ Resolution
  □ Outcome documented
  □ Lessons captured
  □ Process improvements identified
  □ Stakeholders informed

Anti-Patterns

• Alert fatigue — Too many low-priority alerts mask real risks
• Single signal reliance — Missing multi-factor risk patterns
• Slow escalation — Waiting too long to involve leadership
• No playbooks — Ad-hoc response to predictable situations
• Discount-first saves — Training customers to threaten churn
• Ignoring small accounts — Risk exists at all ARR levels
• No documentation — Same mistakes repeated
• Hero culture — Depending on individuals vs. process

Customer Segmentation & Tier Scoring

Impact: MEDIUM-HIGH

Not all customers are equal — and treating them equally means over-investing in some and under-investing in others. Effective segmentation enables right-sized engagement models, focused resources, and segment-specific success strategies. Tier scoring determines service levels.

The Segmentation Framework

┌──────────────────────────────────────────────────────────────────┐
│                   SEGMENTATION DIMENSIONS                        │
├──────────────────────────────────────────────────────────────────┤
│                                                                  │
│  VALUE                    HEALTH                  POTENTIAL      │
│  (Current Worth)          (Current State)         (Future Worth) │
│                                                                  │
│  • ARR / MRR              • Health score          • Growth rate  │
│  • Lifetime value         • Engagement level      • Expansion    │
│  • Contract length        • Risk tier               capacity     │
│  • Payment history        • NPS/Sentiment         • Strategic    │
│                                                                  │
│                          ↓                                       │
│                                                                  │
│                 CUSTOMER TIER ASSIGNMENT                         │
│                                                                  │
│                 Enterprise │ Growth │ Scale │ Tech-touch        │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Value-Based Segmentation

Tier Scoring Model

Customer Tier Score Calculation

TIER SCORE = (Value Score × 0.4) + (Potential Score × 0.35) + (Strategic Score × 0.25)

Value Score Components (0-100):
├── ARR percentile (50%)
├── Contract length (25%)
└── Payment reliability (25%)

Potential Score Components (0-100):
├── Growth trajectory (40%)
├── Seat expansion capacity (30%)
└── Product fit depth (30%)

Strategic Score Components (0-100):
├── Brand recognition (35%)
├── Reference potential (35%)
└── Market influence (30%)

Tier Assignment:
├── Tier 1 (Enterprise): Score 80-100
├── Tier 2 (Growth):     Score 60-79
├── Tier 3 (Scale):      Score 40-59
└── Tier 4 (Tech-touch): Score 0-39

Good Tier Assignment

Customer Tier Assessment: TechCorp Inc.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

VALUE SCORE: 72/100
├── ARR: $65,000 (68th percentile)        → 68 pts
├── Contract: 2-year agreement            → 80 pts
└── Payment: Always on time               → 100 pts
Weighted: 72

POTENTIAL SCORE: 85/100
├── Growth: 25% user growth last year     → 90 pts
├── Expansion: Using 40% of seats         → 75 pts
└── Product fit: 8/10 use cases match     → 80 pts
Weighted: 85

STRATEGIC SCORE: 68/100
├── Brand: Known regional player          → 60 pts
├── Reference: Willing, used once         → 75 pts
└── Influence: 500 LinkedIn followers     → 70 pts
Weighted: 68

TOTAL TIER SCORE: 75.3

Tier Assignment: TIER 2 (Growth)
Engagement Model: Pooled CSM with proactive touchpoints
Rationale: Strong potential for expansion, moderate current value

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Bad Tier Assignment

Customer Tier: Enterprise

Reason: They asked for a dedicated CSM.

Problems:
✗ No objective criteria
✗ Based on customer request, not value
✗ No scoring methodology
✗ No potential assessment
✗ No strategic consideration
✗ Will lead to misallocated resources

Behavioral Segmentation

Segment-Specific Success Strategies

ENTERPRISE SEGMENT ($100K+ ARR)

Engagement Model:
├── Named Strategic CSM (1:10-15 ratio)
├── Dedicated Executive Sponsor
├── Quarterly Business Reviews
├── Annual Strategic Planning
└── Direct access to product leadership

Success Activities:
├── Monthly strategic check-ins
├── Bi-weekly operational reviews
├── Custom success plans
├── Early access to roadmap
└── Executive-level escalation path

Metrics Focus:
├── Value realization / ROI
├── Stakeholder satisfaction
├── Strategic alignment
├── Expansion pipeline
└── Reference/advocacy activity

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

TECH-TOUCH SEGMENT (<$5K ARR)

Engagement Model:
├── Automated, digital-first
├── Community-based support
├── Self-service resources
└── Exception-based human touch

Success Activities:
├── Automated onboarding sequences
├── In-app guidance and tutorials
├── Community forum engagement
├── Triggered outreach (risk, expansion)
└── Scaled webinars and office hours

Metrics Focus:
├── Activation rate
├── Feature adoption
├── Support ticket volume
├── Self-service resolution
└── Upgrade conversion rate

Customer Matrix: Value vs Health

                    HIGH VALUE
                        │
    ┌───────────────────┼───────────────────┐
    │                   │                   │
    │   AT-RISK         │   CHAMPIONS       │
    │   HIGH VALUE      │   HIGH VALUE      │
    │                   │                   │
    │   Strategy:       │   Strategy:       │
    │   Save & retain   │   Expand & grow   │
    │   Executive focus │   Advocacy focus  │
    │                   │                   │
────┼───────────────────┼───────────────────┼────
LOW │                   │                   │ HIGH
HEALTH                  │                   │ HEALTH
────┼───────────────────┼───────────────────┼────
    │                   │                   │
    │   AT-RISK         │   HEALTHY         │
    │   LOW VALUE       │   LOW VALUE       │
    │                   │                   │
    │   Strategy:       │   Strategy:       │
    │   Evaluate ROI    │   Self-serve      │
    │   Tech-touch save │   Upgrade path    │
    │                   │                   │
    └───────────────────┼───────────────────┘
                        │
                   LOW VALUE

Segment Migration Tracking

Resource Allocation by Segment

Segmentation Dashboard

Segmentation Overview

┌─────────────────────────────────────────────────────────────────┐
│  TIER DISTRIBUTION                                              │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  Tier        Customers    ARR         Health Avg   NRR         │
│  ─────────────────────────────────────────────────────────────  │
│  Enterprise      48       $9.8M       78           125%        │
│  Mid-Market     156       $8.2M       72           112%        │
│  SMB            187       $4.8M       68           98%         │
│  Tech-Touch     412       $2.2M       62           92%         │
│                                                                 │
│  Total          803       $25.0M      68           108%        │
│                                                                 │
├─────────────────────────────────────────────────────────────────┤
│  TIER MOVEMENT (Last Quarter)                                   │
│                                                                 │
│  ↑ Upgraded:    34 customers (+$1.2M ARR impact)               │
│  ↓ Downgraded:  12 customers (-$380K ARR impact)               │
│  → Churned:     28 customers (-$520K ARR impact)               │
│  ★ New:         67 customers (+$890K ARR impact)               │
│                                                                 │
├─────────────────────────────────────────────────────────────────┤
│  TIER-SPECIFIC ALERTS                                           │
│                                                                 │
│  Enterprise: 2 accounts at-risk (need exec attention)          │
│  Mid-Market: 8 accounts approaching Enterprise threshold       │
│  SMB: 15 accounts declining, intervention needed               │
│  Tech-Touch: Upgrade candidates identified (12 accounts)       │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Segmentation Implementation Checklist

□ Segment Definition
  □ Clear criteria for each tier
  □ Scoring methodology documented
  □ Thresholds validated against data
  □ Edge case handling defined

□ Data Requirements
  □ Value metrics available
  □ Potential indicators tracked
  □ Strategic scoring inputs defined
  □ Automated calculation possible

□ Engagement Models
  □ CSM ratios defined per tier
  □ Touchpoint cadence specified
  □ Resource allocation approved
  □ Escalation paths documented

□ Migration Rules
  □ Upgrade triggers defined
  □ Downgrade criteria specified
  □ Review process for changes
  □ Customer communication plan

□ Technology Setup
  □ Tier field in CRM
  □ Automated tier calculation
  □ CSM assignment automation
  □ Reporting by segment

□ Team Readiness
  □ CSMs understand segment strategies
  □ Playbooks exist per segment
  □ Training completed
  □ Metrics tracked by segment

Anti-Patterns

• ARR-only tiers — Ignoring potential and strategic value
• Manual assignment — Subjective, inconsistent tiering
• Static segmentation — Not updating as customers change
• One-size engagement — Same model regardless of tier
• Segment leakage — Enterprise service for SMB pricing
• Ignoring potential — Only looking at current value
• No migration path — Customers stuck in initial tier
• Resource mismatch — High-touch for low-value, or vice versa