How-to

Analytics

Read call outcomes, sentiment trends, and drill into individual calls for transcripts and recordings.

Every call leaves a structured record: outcome, duration, transcript, recording, sentiment score. The Analytics section of the dashboard is where you query and visualise these.

What to look at first

  1. Outcome distribution — are calls ending the way you expect? A spike in “no answer” or “abruptly ended” points at a calling-hours or template issue.
  2. Completion rate — percentage of attempted calls that got a meaningful outcome. Low completion over a sustained window is an operational signal (bad numbers, provider issue, template bug).
  3. Sentiment — tracked per call. Sudden drops correlate with template regressions; investigate with the call-level transcripts.

Filter and drill down

  • By template — compare variants against each other.
  • By merchant — only relevant for reseller users.
  • By date range — day, week, month. Default is last 7 days.
  • By outcome — find every “rescheduled” or “transferred to human” call.

Click any row to open the call detail view: full transcript, audio recording, LLM trace (for developers), and the outcome reason.

Export

Every query supports CSV export. Use it for board reports or feeding into your BI tool. Rate-limited server-side — heavy exports queue.

Analytics on the API side

The dashboard is a UI on top of POST /analytics/query. For custom dashboards or scheduled reports, see Analytics reference and run the same queries your team uses via API.

Build a weekly habit

Pick three metrics that matter to your business (completion rate, transfer rate, positive-sentiment rate) and review them every Monday. Trends reveal problems earlier than any single call does.

Next steps

Was this helpful?