Read voice sequence metrics

Open a sequence from Sequences and select Metrics. Choose the period, timezone, mode, and publication before comparing results.

Metrics group runs by their admission time. A call admitted before midnight that finishes after midnight updates the earlier period. Later status updates can change a period you already viewed.

The default mode is production. Select test mode to inspect real test calls. Previews do not create calls or contribute to these metrics. Check mode filters when the Runs list and Metrics totals differ.

Success and failure rates exclude canceled and unfinished runs. A rate with no applicable denominator is unavailable.

The p50 duration is the median: half of the measured completed runs finish within that time. The p95 duration is the time within which 95% finish. A long conversation or an intentional pause can increase elapsed duration without a processing fault.

Publication breakdowns separate runs by the version they used. Draft tests have no published version. Traffic and input differences can affect a comparison even when a workflow change is small.

Technical failure codes describe failed runs with a retained technical cause. Exit reasons describe authored business results. A handled Dial failure can still lead to a successful workflow, and a configured failed ending need not have a technical error code.

Use Runs to inspect examples before changing the workflow. Use the call log for answer and telephone outcomes; sequence success is a different measure.

Reporting is provisional. Unavailable values and gaps do not mean zero activity, and a retained zero does not establish complete historical coverage. Check coverage before comparing periods.

Run detail can show a newly settled outcome before statistics update. Aggregate facts can remain after detailed history expires or a sequence is deleted. Historical totals cannot recover a missing recording or expired step payload.