playbooks

home / skills / gtmagents / gtm-agents / risk-scoring-framework

risk-scoring-framework skill

/plugins/customer-success/skills/risk-scoring-framework

This skill helps build and tune customer health scores by integrating signals, weighting, and governance to align RevOps, CS, and product.

npx playbooks add skill gtmagents/gtm-agents --skill risk-scoring-framework

Review the files below or copy the command above to add this skill to your agents.

Files (1)
SKILL.md
1.3 KB
---
name: risk-scoring-framework
description: Method for calculating customer health/risk tiers using quantitative
  + qualitative data.
---

# Risk Scoring Framework Skill

## When to Use
- Building or tuning customer health scores that drive CS prioritization.
- Aligning RevOps, CS, and product on what “healthy” vs “at-risk” looks like.
- Auditing why certain segments churn or expand more than others.

## Framework
1. **Signal Inventory** – usage, sentiment, support, commercial, product feedback, exec engagement.
2. **Weighting & Decay** – assign weights per signal, set freshness decay, define negative indicators.
3. **Tier Mapping** – convert scores to tiers (green/yellow/red) with playbook hooks.
4. **Validation Loop** – back-test against churn, expansion, and NPS outcomes.
5. **Governance** – review cadence, owner accountability, and change management process.

## Templates
- Signal catalog spreadsheet with weights and owners.
- Tier thresholds + play mapping sheet.
- Validation report template comparing scores vs outcomes.

## Tips
- Combine structured data with CSM notes or sentiment highlights for context.
- Keep tiers simple (3-4) to avoid confusion; use tags for nuance.
- Pair with `monitor-customer-health` output to auto-highlight risks.

---

Overview

This skill defines a repeatable risk-scoring framework to calculate customer health and risk tiers using a blend of quantitative and qualitative signals. It provides a structured inventory of signals, weighting and decay rules, tier mapping, validation steps, and governance to keep scores actionable. The outcome is a production-ready approach teams can use to prioritize accounts and align cross-functional playbooks.

How this skill works

The framework starts with a signal inventory capturing usage metrics, support interactions, commercial data, product feedback, sentiment, and executive engagement. Each signal is assigned a weight and freshness decay so recent events matter more than stale ones, and negative indicators can reduce scores. Scores are converted into simple tiers (e.g., green/yellow/red) tied to playbook actions. A validation loop back-tests scores against churn, expansion, and NPS, and governance defines cadence and ownership for ongoing tuning.

When to use it

  • Designing or tuning a customer health score to drive CS prioritization
  • Aligning RevOps, CS, and Product on definitions of healthy vs. at-risk accounts
  • Auditing segments that show unusual churn or expansion rates
  • Triggering automated playbooks or alerts for at-risk customers
  • Validating score effectiveness before rollout or after major product changes

Best practices

  • Keep tiers simple (3–4) and use tags for nuance rather than many tiers
  • Maintain a central signal catalog with owners, definitions, and weights
  • Apply decay rules so recent signals have higher influence than old ones
  • Include qualitative inputs (CSM notes, sentiment highlights) to complement structured data
  • Run regular validation against outcomes and adjust weights via a defined governance cadence

Example use cases

  • Prioritizing CSM time by highlighting accounts in the red tier for immediate outreach
  • Automating renewal and expansion playbooks based on tier changes
  • Investigating why a cohort has higher churn by comparing signal contributions
  • Reporting health distribution to executives with clear playbook triggers
  • Feeding risk tiers into monitoring tools to surface hot accounts to RevOps

FAQ

How many signals should I track initially?

Start with a focused set (8–12) covering usage, support, commercial, sentiment, and engagement, then expand as needed.

How often should scores be recalculated?

Recalculate at least daily for active accounts, and real-time or hourly for high-touch enterprise customers.

What if qualitative notes contradict the numerical score?

Surface contradictions as flags for CSM review and consider adding sentiment or recent notes as weighted signals.