AI In-App Support
Deploy an AI-powered support chatbot (Intercom Fin) inside your product to resolve user questions autonomously, reduce human ticket volume, and feed support patterns into churn prediction.
npx gtm-skills add product/retain/ai-support-chatbotOutcome
≥50% engagement; ≥30% AI resolution rate
Leading Indicators
- Chatbot engagement rate
- AI resolution rate
- CSAT for AI-resolved conversations
Instructions
-
Audit Intercom Help Center: verify articles cover top 20 support questions. Write missing articles.
-
Run ai-chatbot-deployment drill: enable Fin AI, connect knowledge sources, create 5 custom answers, configure escalation rules.
-
Instrument PostHog events: chatbot_conversation_started, chatbot_resolved_by_ai, chatbot_escalated_to_human, chatbot_csat_submitted, chatbot_article_suggested.
-
Deploy behind PostHog feature flag to 10-20% of users.
-
Monitor daily for 7 days: engagement, resolution rate, escalation topics.
-
Do not tune for first 3 days. After day 3, add custom answers for dominant escalation topic if needed.
-
Run threshold-engine drill: engagement ≥50%, AI resolution rate ≥30%, CSAT ≥3.5.
-
If PASS, proceed to Baseline. If FAIL, audit top escalation topics and add knowledge base content.
Recommendations
Time
5 hours over 1 week
Play-specific cost
Intercom Essential ($29/seat/mo) + Fin AI (~$50 in resolutions for test cohort)