Skip to main content

Customer Health Model - Playbook

1. Definition​

What it is: A technical and strategic implementation project that creates a multi-dimensional scoring system to measure customer health by combining product usage signals, CSM sentiment, support interactions, and engagement metrics into a unified health score that enables proactive churn prevention.

What it is NOT: Not a Customer Success Platform implementation (that's the tool setup). Not Customer Segmentation (that's account tiering). Not a Renewal Forecasting model (that's revenue prediction). Not NPS/CSAT survey setup (that's one input to health scoring).

2. ICP Value Proposition​

Pain it solves: Customer Success teams lack visibility into which accounts are truly at-risk versus healthy. CSM "gut feel" is inconsistent and doesn't scale. Churn surprises happen because warning signs weren't systematically tracked. Without a data-driven health score, intervention happens too late.

Outcome delivered: A functioning customer health scoring system with clear thresholds (Red/Yellow/Green), automated data collection from product and CRM, dashboards for visibility, and defined playbooks for at-risk intervention. CSMs can prioritize their book of business based on objective signals rather than intuition.

Who owns it: VP of Customer Success or Head of Customer Operations.

3. Implementation Procedure​

Part 1: Discovery & Requirements Definition​

Step 1: Analyze Historical Churn Data​

Step Overview: Review past churn events to identify patterns and signals that preceded customer departures. End state: Documented list of leading indicators correlated with churn in your specific business.

  • Pull list of all churned customers from past 12-24 months
  • Categorize churn reasons (product fit, support issues, champion left, budget, competitor)
  • Identify common behaviors before churn (usage decline, support ticket spikes, missed QBRs)
  • Interview 2-3 CSMs on what they noticed before accounts churned
  • Document 5-10 potential health score signals based on churn analysis

Step 2: Define Health Score Dimensions and Metrics​

Step Overview: Identify the specific metrics that will comprise the health score across multiple dimensions. End state: Finalized list of 5-10 metrics organized by dimension with clear definitions.

  • Define Product dimension metrics (login frequency, feature adoption, license utilization, active user trend)
  • Define Engagement dimension metrics (QBR attendance, email response rate, executive sponsor engagement)
  • Define Support dimension metrics (ticket volume, escalation frequency, time to resolution)
  • Define Financial dimension metrics (payment timeliness, expansion history, renewal timing)
  • Include CSM sentiment as a qualitative override dimension
  • Document data source for each metric (CRM, product analytics, support system)

Step 3: Validate Metrics with Stakeholders​

Step Overview: Review proposed metrics with CS leadership and cross-functional partners to ensure buy-in. End state: Approved metric list with stakeholder alignment on what signals matter.

  • Schedule review meeting with VP Customer Success and CS managers
  • Present proposed metrics with rationale from churn analysis
  • Gather feedback on missing signals or irrelevant metrics
  • Validate data availability with RevOps/Data team
  • Finalize approved metric list with documented rationale

Part 2: Scoring Model Design​

Step 1: Establish Metric Weights​

Step Overview: Assign relative importance weights to each metric based on correlation with retention outcomes. End state: Weighted scoring model with percentages assigned to each metric.

  • Analyze historical correlation between each metric and renewal/churn outcomes
  • Assign weights totaling 100% across all metrics (e.g., product usage 40%, engagement 25%, support 20%, financial 15%)
  • Consider segment-specific weights (enterprise vs SMB may weight differently)
  • Document rationale for each weight assignment
  • Create scoring formula combining weighted metrics

Step 2: Define Scoring Thresholds and Categories​

Step Overview: Establish the score ranges that define Red (at-risk), Yellow (needs attention), and Green (healthy) accounts. End state: Clear threshold definitions with category descriptions.

  • Set numeric thresholds for each category (e.g., 0-40 Red, 41-70 Yellow, 71-100 Green)
  • Define what each category means operationally (Red = immediate intervention required)
  • Document edge cases and override rules (e.g., CSM can override if major event occurred)
  • Consider adding "Black" category for accounts with insufficient data
  • Test thresholds against historical churned accounts to validate accuracy

Step 3: Design Segment-Specific Scoring Rules​

Step Overview: Customize scoring criteria for different customer segments to ensure fair comparison. End state: Segment-specific scoring rules that account for usage patterns by account size.

  • Define customer segments (Enterprise, Mid-Market, SMB, or by product tier)
  • Adjust usage expectations by segment (50% login rate may be healthy for Enterprise, concerning for SMB)
  • Set different metric weights per segment if needed
  • Document lookup windows for each metric (e.g., trailing 30 days vs 90 days)
  • Create segment mapping rules for automated categorization

Part 3: Technical Implementation​

Step 1: Configure Data Integrations​

Step Overview: Set up data flows from source systems into the customer success platform. End state: All required data sources connected and syncing reliably.

  • Connect CRM (Salesforce/HubSpot) to CS platform via native integration
  • Set up product analytics integration (Segment, Amplitude, Mixpanel, or direct API)
  • Configure support system integration (Zendesk, Intercom, Freshdesk)
  • Establish billing/financial data sync from payment system
  • Verify data freshness and sync frequency for each source
  • Document any data transformation or field mapping required

Step 2: Build Health Score in CS Platform​

Step Overview: Configure the health score calculation in Gainsight, ChurnZero, Vitally, or similar platform. End state: Automated health score calculating correctly for all accounts.

  • Create health score measures for each defined metric
  • Configure metric calculations with correct lookback windows
  • Apply weights to each measure per approved model
  • Set up segment-based scoring variations if applicable
  • Configure score refresh frequency (daily recommended)
  • Set up CSM sentiment input field for manual override

Step 3: Create Health Score Dashboard​

Step Overview: Build reporting dashboards for health score visibility across the CS organization. End state: Dashboard showing account health distribution with drill-down capability.

  • Create executive summary view (health distribution pie chart, trend over time)
  • Build CSM-level view filtered by book of business
  • Include health score trend charts for individual accounts
  • Add drill-down to see which metrics are driving low scores
  • Configure alerts for accounts crossing threshold boundaries
  • Set up scheduled dashboard delivery to CS leadership

Part 4: Validation & Testing​

Step 1: Test Score Accuracy Against Historical Data​

Step Overview: Validate the health score model by checking if it correctly identifies known churned accounts. End state: Validated model with documented accuracy metrics.

  • Run health scores retrospectively against accounts that churned 6+ months ago
  • Calculate false positive rate (Green accounts that churned)
  • Calculate false negative rate (Red accounts that renewed/expanded)
  • Adjust thresholds or weights if accuracy is below 70%
  • Document model accuracy baseline for future comparison
  • Identify any segments where model performs poorly

Step 2: Conduct Pilot with CSM Team​

Step Overview: Run pilot period with subset of CSMs to validate score usefulness and gather feedback. End state: Pilot complete with documented feedback and refinements.

  • Select 2-3 CSMs for pilot program (mix of tenure and segment coverage)
  • Review their top 10 accounts by health score (high and low)
  • Validate scores match CSM intuition and known account situations
  • Gather feedback on missing signals or incorrect assessments
  • Refine model based on pilot learnings
  • Document changes made and rationale

Part 5: Rollout & Enablement​

Step 1: Develop Intervention Playbooks​

Step Overview: Create standardized response playbooks for each health score category. End state: Documented playbooks defining actions for Red, Yellow, and Green accounts.

  • Define Red account intervention steps (executive escalation, QBR scheduling, product review)
  • Define Yellow account monitoring cadence and proactive touchpoints
  • Define Green account expansion opportunity identification
  • Create escalation path for accounts declining into Red status
  • Document SLA for CSM response to score changes
  • Build playbook templates in CS platform if supported

Step 2: Train Customer Success Team​

Step Overview: Enable the CS team on health score interpretation and usage. End state: Team trained and confident using health scores in daily workflow.

  • Schedule training session (45-60 minutes) with full CS team
  • Explain each metric, weight, and threshold rationale
  • Demonstrate dashboard navigation and account drill-down
  • Review intervention playbooks and expected actions
  • Practice with sample accounts (what would you do for this Red account?)
  • Create quick-reference guide for ongoing use

Step 3: Hand Off to Client​

Step Overview: Transfer ownership and establish ongoing governance process. End state: Client self-sufficient with admin access and quarterly review process defined.

  • Transfer admin access to CS Operations or RevOps owner
  • Deliver configuration documentation (metrics, weights, thresholds, data sources)
  • Establish quarterly health score review cadence to assess accuracy
  • Define process for requesting metric or threshold changes
  • Schedule 30-day check-in to review adoption and address questions
  • Close out project with success metrics documentation

4. Dependencies & Inputs​

What must exist before starting:

  • Customer Success Platform in place (Gainsight, ChurnZero, Vitally, Catalyst, Totango, Planhat)
  • CRM with customer account data (Salesforce or HubSpot)
  • At least 6-12 months of customer data (for churn analysis and validation)
  • Product analytics tracking in place (or ability to implement)
  • Defined customer segments or tiers

What client must provide:

  • Access to CS platform with admin permissions
  • Access to CRM with reporting permissions
  • Historical churn data with reason codes if available
  • Product usage data or analytics platform access
  • 2-3 hours of stakeholder time for requirements and validation sessions
  • CSM team availability for training (45-60 minutes)

5. Common Pitfalls​

  • Overcomplicating the model with too many metrics: Starting with 20+ metrics makes the score confusing and hard to trust. Teams won't use what they can't explain. Mitigation: Start with 5-7 high-impact metrics. Add complexity only after proving the base model works.

  • Using the same thresholds for all customer segments: A 50% product usage rate might be healthy for Enterprise (heavy users log in weekly) but concerning for SMB (should be daily). Mitigation: Benchmark usage against customer size/segment and set segment-specific thresholds.

  • Relying too heavily on CSM sentiment or a single metric: Manual sentiment alone is inconsistent across CSMs and doesn't scale. Single-metric scores miss important signals. Mitigation: Balance quantitative data (usage, support) with qualitative input (sentiment) and require multiple signals to trigger Red status.

  • Setting and forgetting the health score model: Customer behavior and product features change. A model built in Q1 may be inaccurate by Q4. Mitigation: Schedule quarterly health score reviews to analyze false positives/negatives and adjust weights or thresholds accordingly.

6. Success Metrics​

  • Leading Indicator: CSMs are actively using health scores in weekly workflow (measured by dashboard views and score-based actions taken within 14 days of launch)
  • Lagging Indicator: Reduction in "surprise" churn (accounts that churned without being flagged Red 60+ days prior) by 50% within first 6 months