Analytics & AI performance

Analytics (on the dashboard or per property) show how many messages you’re getting, how many the AI is handling, and how often it escalates or needs review.

What’s measured

• Total messages — Incoming and outgoing in the selected period (e.g. 7d, 30d, 90d).
• AI messages — Replies generated (and usually sent) by the AI.
• Host messages — Replies sent by you (manual or after editing an AI suggestion).
• AI reply rate — Share of outgoing messages that were AI-generated (vs manual).
• Escalations — Count of conversations where the AI escalated (refund, complaint, etc.) and did not send.
• Pending review — Count where the AI suggested a reply but waited for your approval.
• Intent distribution — Breakdown of what the AI thought the guest was asking (check-in, WiFi, local tips, etc.), so you can tune FAQ and escalation.

Scope

• Global — All your properties combined.
• Per property — Select one property to see its metrics and intents. Useful to see which properties get the most messages or need the most review.

Using the data

• Low AI reply rate — Maybe the AI is off for some properties, or many messages are escalated. Check AI Config and escalation settings.
• High escalation — Consider whether you want to add more FAQ or adjust escalation intents so the AI can answer more without escalating.
• Intents — If one intent (e.g. “check-in question”) dominates, ensure your check-in rules and FAQ are complete so the AI stays confident.

Next: Messages not syncing — Checklist when messages don’t appear.