Skip to content
VeraFrame Docs

Analytics

The Analytics tab in the Admin dashboard gives you a tenant-wide view of how VeraFrame is being used and how AI output is performing against your source material.

Overview metrics

The top of the analytics page shows rolling counts over a 30-day and 90-day window:

  • Total validations — how many generations, audits, asks, and JSON API calls were run.
  • Mode breakdown — the split across Generate / Audit / Ask / JSON API.
  • Issue distribution — how many validations were high / medium / low confidence.
  • Accuracy rate — the percentage of verified claims across all validations.

These are useful for “is VeraFrame being adopted?” and “is our AI output getting better?” questions.

Validation history

The Validations tab lists every individual validation run in the tenant, with filters for:

  • Confidence level — high / medium / low
  • Mode — Generate / Audit / Ask / JSON API
  • Channel — UI / REST API / Integration
  • User — filter by who ran the validation
  • Date range

Click any row to open the validation detail: the original request, the AI output, the full trust report, and (if applicable) the corrections made to the output. This detail view is the same one that reviewers use in the approve/reject workflow.

Export

Two export options are available on tenants with the export_reports feature:

  • CSV export — a flat table of validations with metadata (timestamp, user, mode, confidence, issue count, reviewed state). Useful for importing into BI tools or spreadsheets.
  • JSON export — the full structured data including trust reports and corrections.

Exports cover only the rows currently filtered in the UI, so you can produce targeted data extracts (for example, “all low-confidence validations from last month for the legal team”).

Audit-specific exports

For tenants with audit_trail enabled there is a separate Audit export in the Compliance tab that produces a structured, auditor-ready export. The analytics exports above are for internal reporting; the audit export is formatted for regulatory review. See Audit trail.

Typical uses

  • Sprint retro. Show the team the drop in mismatch rate after a sprint of improving source material.
  • Vendor oversight. When you audit a third-party chatbot via Chatbot Audit mode, track the vendor’s accuracy over time in analytics.
  • Compliance reporting. Monthly or quarterly summary of what was generated, what was reviewed, and where issues clustered.