Support Ticket Churn Signals
Classify support tickets by category, severity, and sentiment; correlate ticket patterns with churn; score accounts for churn risk; trigger tiered CS interventions for at-risk accounts.
npx gtm-skills add product/retain/support-issue-trackingOutcome
>=1 ticket pattern with >=2x churn lift validated against 40+ accounts
Leading Indicators
- Churn-signal lift (>=2x for at least 1 signal)
- Ticket classification accuracy (>=80%)
- Accounts analyzed (>=20 churned + >=20 retained)
Instructions
-
Export 90 days of Intercom support tickets via API.
-
Classify each ticket by category, severity, and sentiment using Claude Haiku.
-
Spot-check 10 random classifications for >=80% accuracy.
-
Identify >=20 churned and >=20 retained accounts from CRM.
-
Aggregate ticket data per account: volume, categories, severity, CSAT, repeat issues.
-
Calculate signal lift for each ticket pattern: churned rate / retained rate.
-
Rank signals by lift. Identify patterns with >=2x lift.
-
If >=1 signal at 2x+, document top 3 signals with coverage and false positive rate.
-
Proceed to Baseline or iterate if no signal exceeds 1.5x.
Recommendations
Time
5 hours over 1 week
Play-specific cost
<$1 (LLM classification of historical tickets)