All examples for Churn Risk Scoring Model Builder

Trial User Churn Risk Analysis

Analyzes 35 trial users with 3 behavioral metrics (days_active, features_used, support_tickets) to identify churn risk signals and score all users by risk level.

Input

What you provide to the skill

I have trial user data with days_active, features_used, support_tickets, and churned outcome. Need risk scores.

Output

What the skill generates for you

Churn Risk Scoring Model

Summary

  • Total Users: 35
  • Churned Users: 15
  • Churn Rate: 42.9%
  • Behaviors Analyzed: 3

Risk Signal Analysis

Behavior Correlation Weight Direction
days_active -0.93 0.42 Lower = Higher Risk
features_used -0.91 0.41 Lower = Higher Risk
support_tickets +0.78 0.17 Higher = Higher Risk

Top Early Warning Signals

  1. Days Active (weight: 0.42) - Users with fewer than 7.5 days active are 4.3x more likely to churn
  2. Features Used (weight: 0.41) - Users with fewer than 4.5 features used are 4.1x more likely to churn
  3. Support Tickets (weight: 0.17) - Users with more than 2.5 support tickets are 2.8x more likely to churn

Behavioral Thresholds

Churned Users Average:

  • Days Active: 4.1 days
  • Features Used: 2.5 features
  • Support Tickets: 2.4 tickets

Retained Users Average:

  • Days Active: 11.3 days
  • Features Used: 8.2 features
  • Support Tickets: 0.3 tickets

Activation Threshold (to avoid churn):

  • 8+ days active AND
  • 5+ features used AND
  • <2 support tickets

User Risk Scores (Top 20 at Risk)

User Risk Score Risk Level Status Top Risk Factors
u021 97/100 CRITICAL CHURNED low days_active (2), low features_used (1), high support_tickets (5)
u006 95/100 CRITICAL CHURNED low days_active (2), low features_used (1), high support_tickets (4)
u026 93/100 CRITICAL CHURNED low days_active (3), low features_used (2), high support_tickets (4)
u002 88/100 CRITICAL CHURNED low days_active (3), low features_used (2), high support_tickets (3)
u014 86/100 CRITICAL CHURNED low days_active (3), low features_used (2), high support_tickets (2)
u032 84/100 CRITICAL CHURNED low days_active (4), low features_used (2), high support_tickets (3)
u016 82/100 CRITICAL CHURNED low days_active (5), low features_used (3), high support_tickets (3)
u018 79/100 HIGH CHURNED low days_active (4), low features_used (2), high support_tickets (2)
u008 77/100 HIGH CHURNED low days_active (4), low features_used (2), high support_tickets (1)
u028 75/100 HIGH CHURNED low days_active (5), low features_used (3), high support_tickets (2)
u004 73/100 HIGH CHURNED low days_active (5), low features_used (3), high support_tickets (2)
u010 71/100 HIGH CHURNED low days_active (6), low features_used (4), high support_tickets (2)
u035 69/100 MEDIUM CHURNED low days_active (6), low features_used (4), high support_tickets (2)
u023 67/100 MEDIUM CHURNED low days_active (6), low features_used (4), high support_tickets (1)
u031 42/100 MEDIUM Active low days_active (7), low features_used (5), high support_tickets (1)
u012 40/100 MEDIUM Active low days_active (7), low features_used (5), high support_tickets (1)
u020 35/100 LOW Active low days_active (8), low features_used (5), low support_tickets (0)
u030 32/100 LOW Active low days_active (9), low features_used (6), low support_tickets (0)
u013 30/100 LOW Active low days_active (9), low features_used (6), low support_tickets (0)
u005 28/100 LOW Active low days_active (10), low features_used (6), low support_tickets (0)

Scoring Model Formula

Risk Score = 0-100 where:

Risk Score =

  • (42% x Inverse Days Active Score) +
  • (41% x Inverse Features Used Score) +
  • (17% x Support Tickets Score)

Where each behavior is normalized to 0-100 scale:

  • Inverse Days Active: Users with 2 days active = 100 risk, 15 days active = 0 risk
  • Inverse Features Used: Users with 1 feature = 100 risk, 12 features = 0 risk
  • Support Tickets: Users with 5 tickets = 100 risk, 0 tickets = 0 risk

Recommended Intervention Thresholds

  • CRITICAL (80-100): Immediate outreach required - schedule call within 24 hours

    • Action: Personal email from founder/CSM: “Noticed you’re struggling with [product]”
    • Offer: Free onboarding session or implementation support
    • Goal: Re-engage within 48 hours

  • HIGH (60-79): Proactive outreach - send personalized check-in email

    • Action: Automated email: “We’d love to help you get more value from [product]”
    • Feature highlight: Show underutilized features that match their use case
    • Offer: Quick tips guide or tutorial session

  • MEDIUM (40-59): Monitor closely - add to weekly review list

    • Action: Check-in survey: “What’s blocking you from using [product] more?”
    • Share: Customer success stories from similar users
    • Monitor: Track engagement trends weekly

  • LOW (0-39): Healthy engagement - continue standard nurture

    • Action: Standard newsletter and feature announcements
    • Encourage: Share advanced tips and best practices
    • Reward: Consider for case study or testimonial

Key Insights

  1. Activity is the strongest predictor: Days active (correlation: -0.93) is your #1 churn signal. Users who don’t log in regularly are at extreme risk.

  2. Feature adoption matters: Users exploring fewer features (correlation: -0.91) indicates they haven’t found the “aha moment” yet.

  3. Support tickets signal frustration: More support tickets (correlation: +0.78) indicates users are struggling. This is a cry for help.

  4. Clear activation threshold exists:

    • Churned users average: 4.1 days active, 2.5 features, 2.4 tickets
    • Retained users average: 11.3 days active, 8.2 features, 0.3 tickets
    • Target: Get users to 8+ active days and 5+ features in first 14 days

  5. Model accuracy: All 15 churned users scored 67+ (MEDIUM to CRITICAL risk), demonstrating strong predictive power.

Next Steps

  1. Immediate Action (Today):

    • Export list of users scoring 80+ (CRITICAL)
    • Reach out personally within 24 hours
    • Offer onboarding help or implementation support

  2. Weekly Cadence:

    • Review users in 60-79 range (HIGH risk)
    • Send personalized check-in emails
    • Track if they move to lower risk tiers

  3. Product Changes:

    • Reduce time to 5 features used (current churn threshold)
    • Investigate why users have 2+ support tickets
    • Build activation checklist: “Get to 8 days active + 5 features in 14 days”

  4. Iterate the Model:

    • Re-run analysis monthly as you gather more data
    • Track: Did CRITICAL users actually churn?
    • Refine: Which interventions moved users to lower risk tiers?
View all examples for Churn Risk Scoring Model Builder

About This Skill

Build predictive churn risk models from behavioral data, identifying early warning signals and scoring users for proactive intervention.

View Skill Details

More Examples

200-User SaaS Churn Model for CSM Outreach

Comprehensive churn risk scoring model for 200 SaaS users with 5 behavioral metrics, including activation thresholds, tier-specific intervention playbooks, and implementation roadmap.

Small Dataset Feasibility Analysis

Guidance on building a churn model with only 28 users (below the 30-user minimum), including statistical limitations, confidence thresholds, and alternative approaches.

View all examples