Track Fin AI Escalations in Intercom
Every Fin escalation is a data point. It tells you that a customer's issue was outside Fin's current capability. The pattern of escalations tells you where to invest in Fin content next.
The manual review approach
- 1Export Fin escalation data from Intercom analytics.
- 2Categorize escalation reasons manually.
- 3Prioritize content improvements based on frequency.
The problem: Manual categorization of escalation reasons requires reading transcripts, which is time-consuming and inconsistent.
With Supportman
Fin escalations post to Slack as they happen, with the conversation link. Your team can tag them in Slack threads to build a qualitative pattern before committing to content work.
- Enable Fin escalation tracking in Supportman.
- Each escalation appears in Slack with conversation context.
- Team leads tag escalation reasons in the Slack thread to identify patterns.
How do I know if an escalation was because Fin could not answer or because the customer requested a human?
Intercom distinguishes between Fin routing to human (AI escalation) and customer-initiated handoff. Supportman surfaces the Intercom-provided reason in the Slack message.
Can I use escalation patterns to train Fin?
The content improvements to Fin (updating articles, adding Q&A pairs) are done in Intercom. Supportman helps you identify which gaps to prioritize by making escalations visible.