Intelligence & Reports
Reports glossary
A plain-language reference for the terms that appear across HabariChat Reports. Learn them once here.
Glossary
| Term | Meaning |
|---|---|
| AI resolution % | Outgoing replies the AI sent without human edits ÷ all outgoing replies. |
| Auto-handled rate | Replies approved without any edit ÷ all reviewed replies. |
| CSAT | Customer Satisfaction; average of customer ratings on a 1–5 scale. |
| AI self-rating / confidence | The AI's own 1–5 rating of its answer. |
| Escalation | A conversation handed from the AI to a human. |
| SLA | Service Level Agreement; the response-time promises tracked at 1h / 4h / 24h, and the 24h escalation deadline. |
| Median | The middle value; resistant to a few extreme outliers. |
| Percentile (p50–p99) | "X% of cases were at least this fast"; p95 = the slow tail. |
| Retrieval | The AI fetching relevant chunks from your documents to answer a question. |
| Chunk | A small searchable piece a document is split into for the AI. |
| Zero-result query | A search that matched no document content. |
| Freshness (Current/Aging/Stale) | Document age buckets: ≤7 days / 8–30 days / >30 days. |
| Delivery % / Read % | For campaigns: delivered ÷ sent, and read ÷ delivered. |
| Opt-in / opt-out | Whether a contact has consented to receive marketing. |
| Token | The unit AI usage and cost are measured in. |
| Snapshot | A once-a-day capture of account totals, used for long-term trends. |