Connect to IntercomCompare CSAT for AI vs Human Conversations in Intercom
One of the most important questions when deploying Fin AI is whether the customer experience is as good as human handling — or better. CSAT by conversation type is the cleanest answer to that question.
The manual comparison approach
- 1Export CSAT data from Intercom with conversation type tags.
- 2Separate Fin-handled from human-handled conversations.
- 3Calculate average CSAT for each group.
- 4Compare and draw conclusions.
The problem: This analysis requires either a BI tool or careful spreadsheet work. Most teams never do it consistently, so they have no data to validate Fin's contribution.
With Supportman
CSAT alerts in Supportman include the agent name — Fin appears as the agent for AI-handled conversations. Over time, you can see in your Slack channel history which conversations were Fin-handled and how they rated.
- All CSAT ratings appear in Slack tagged with the handling agent (human or Fin).
- Filter your Slack channel history by "Fin" to see AI-specific ratings.
- Use Intercom Reports for the statistical aggregate if you need formal analysis.
Does Intercom attribute CSAT ratings to Fin AI separately?
Yes — Intercom's CSAT reports can be filtered by whether the conversation was fully handled by Fin, partially handled, or fully human. Supportman surfaces the same attribution in Slack.
What if Fin CSAT is lower than human CSAT — should I reduce Fin deflection?
Not necessarily. A small gap is expected — Fin handles different conversation types than humans. A large gap (>0.5 stars) usually indicates content gaps that can be fixed. Outright removal of Fin deflection is rarely the right answer.