Voice Agent
Get started

Understand your first evaluation

Read a completed run from its suite summary down to metric-level evidence.

After processing, start in Results. The page is headed Test Suites and gives you the broadest view before you inspect individual recordings.

Read the suite first

Open the suite created for your upload. Its detail page shows:

  • Total recordings
  • Average score
  • Status
  • Aggregated performance metrics

These help you understand the group before focusing on one call.

Open the individual run

In Recording breakdown, select Open recording. The run detail includes:

  • Overall Score
  • Metrics Breakdown
  • Transcript
  • Call Recording
  • Metric-detail evidence

Use the transcript and audio to check the context behind a score. Evidence is there to make the result explainable, not just to produce a number.

Understand missing results

A metric may be:

  • Eligible — it can be evaluated from the available call and setup.
  • Needs setup — it requires context or configuration that is not currently available.
  • Not scored — the metric could not produce a result from the available, applicable evidence or setup.

A completed run with some missing metrics is not necessarily a processing failure. Check the metric’s status and evidence before deciding whether to improve call context, agent setup, or the recording itself.

Continue from the finding

Use what you learn to choose the correct next action. When a newly available or changed metric should be applied to this same call, re-evaluate the existing run. When the agent, scenario, or conversation must change, create a fresh comparable run instead.

Learn about testing modes

On this page