Skip to navigation

get_ada_metric

Retrieves performance metrics for the AI Agent. To get conversation summaries, use get_conversations with SUMMARY detail.

Example prompts

  • “What’s our CSAT this week?”
  • “Show my AR and volume for the last 7 days.”
  • “How is my AR this week compared to last week?”
  • “What’s our containment rate trend for the last 30 days?”
  • “Which API tools were slowest last week?”

Parameters

ParameterTypeRequiredDescription
metric_typestringYesThe metric to retrieve. See supported values below.
start_datestringYesStart date in YYYY-MM-DD format.
end_datestringYesEnd date in YYYY-MM-DD format.
filtersarrayNoArray of filter objects from get_available_filters. Each filter has type, operator, and value fields.

Supported metric types

metric_type valueDescription
resolution_ratePercentage (0–100) of automatically resolved conversations.
csat_ratePercentage (0–100) customer satisfaction across all surveys.
csat_rate_aiPercentage (0–100) customer satisfaction for surveys attributed to the AI Agent. A survey is attributed to the AI Agent when no human agent joined the conversation. This includes a conversation that was handed off but that no human agent joined. Email surveys are always attributed to the AI Agent. Matches the AI Agent Score in the Customer satisfaction score report. null when no survey matches.
csat_rate_humanPercentage (0–100) customer satisfaction for surveys attributed to a human agent who joined the conversation. A colleague who joined a conversation that the AI Agent kept does not count. On an Influx handoff, a handoff message alone attributes the survey to a human agent. Matches the Human Agent Score in the Customer satisfaction score report. null when no survey matches.
conversation_volume_engagedCount of conversations where the customer actively engaged.
conversation_volume_openedCount of conversations where a greeting was presented, including those without active engagement.
containment_ratePercentage (0–100) of conversations not escalated to a human agent.
containment_volumeCount of conversations not escalated to a human agent.
avg_handle_timeAverage time in seconds customers spent with the AI Agent for contained conversations.
avg_handle_time_agentAverage time in seconds customers spent with live agents after escalation.
api_tool_performanceOne row per API tool with its call count, error rate, and response time. Excludes code tools and MCP tools. Use the ACTIONID filter for one API tool.

Response

Returns an object with metric_type and value. When csat_rate comes with the HANDOFF filter, the object also has a note that names csat_rate_ai and csat_rate_human.

Dates are in UTC. The dashboard reports use the time zone of the viewer, so a result can differ from the dashboard for a viewer outside UTC.

When using api_tool_performance, each row includes:

FieldDescription
tool_idUnique identifier for the API tool.
tool_nameName of the API tool.
conversationsCount of conversations that used the API tool.
api_callsCount of API calls the API tool made.
error_countCount of API calls that did not succeed.
error_ratePercentage (0–100) of API calls that did not succeed.
timed_api_callsCount of API calls that recorded a response time.
median_response_time_msTime in milliseconds within which half of the timed calls got an answer. null when no call in the range recorded a response time.
p90_response_time_msTime in milliseconds that 1 in 10 timed calls took longer than. null when no call in the range recorded a response time.

Response times are available from September 22, 2026 onward.