Measure

Analytics dashboards

The Overview tab charts trends across many calls, with widget fields, starter dashboards and the source filter.

The Overview tab is a dashboard grid of charts scoped to this agent. It answers the questions a single transcript cannot: how busy is the agent, how long are calls running, how often do they succeed, and how fast is it responding. A source filter on the control bar switches the grid between production, test, eval and all calls. To read one call instead, see Calls.

How a widget is built

Each widget is a chart built from one data field, bucketed by a group and reduced by an aggregation.

ChoiceOptions
Data fieldDuration, success, outcome, sentiment, latency, topic, call-ended reason, and others.
GroupDay, week, hour, agent, outcome, sentiment, topic, and others.
AggregationCount, average, sum, median (p50), p95.

Chart types include line, bar, stacked bar, pie, KPI tiles and lists.

Data field   latency
Group        day
Aggregation  median (p50)
Chart        line

Starter dashboards

Add a widget yourself, or apply a starter dashboard that ships a themed set in one click. For example:

  • Operations: call volume over time, by hour and by agent, plus average duration and top topics.
  • Quality and outcomes: success rate and its trend, outcome mix, sentiment, drop-off points and review coverage.
  • Performance and latency: median (p50) and p95 latency with average call duration over time.
  • Executive at-a-glance: the headline numbers (total calls, average duration, median latency) with a weekly trend.

Where the latency numbers come from

Response latency and time-to-first-audio are measured by the runtime on every call. Latency surfaces as a chartable field here (the Performance and latency dashboard plots p50 and p95). The full per-stage waterfall is on each call's detail page and the Evals scoring views.

On this page