Search Agent query examples

This is for:

System Administrator Business Analyst
In this article

You can retrieve Search Agent data using SQL queries to build your own reports and analyses. Search Agent data is stored in the ANSWERING schema of the CORE data model in Snowflake. You can access the schema through a reader account or a data share.

Query examples

The following sample queries are a starting point for querying Search Agent data.

Note

The sample queries retrieve the QUERY_EXPRESSION column, which lives in the INSIGHTS view, not in ANSWERS. Since the two views reside in different schemas, the JOIN clause is required to combine them. The JOIN keys are ANSWERS.ANSWER_ID = INSIGHTS.INSIGHT_ID and ANSWERS.ANSWER_DATE = INSIGHTS.INSIGHT_DATE, partitioned by ORGANIZATION_ID.

Retrieve answers with their query expression

Use this query to retrieve Search Agent answers alongside the query that triggered them. This is the most common starting point for Search Agent reporting.

SELECT
    A.ORGANIZATION_ID
    , A.CONVERSATION_ID
    , A.ANSWER_XID
    , I.QUERY_EXPRESSION
    , A.IS_QUERY_ANSWERED
    , A.ANSWER
FROM COVEO_CORE_MODEL_V001.COMMON.INSIGHTS I
JOIN COVEO_CORE_MODEL_V001.ANSWERING.ANSWERS A
    ON I.ORGANIZATION_ID = A.ORGANIZATION_ID
    AND I.INSIGHT_ID = A.ANSWER_ID
    AND I.INSIGHT_DATE = A.ANSWER_DATE
WHERE A.ANSWER_SOURCE = 'search_agent'
AND I.INSIGHT_DATE BETWEEN '<START_DATE>' AND '<END_DATE>'

Where you replace:

  • <START_DATE> and <END_DATE> with date values, formatted as YYYY-MM-DD. A conversation can remain active for up to 48 hours, so use a date range that covers the full period of interest.

Retrieve answers with conversation context

Use this query to retrieve all turns from a specific conversation, identified by its CONVERSATION_XID.

SELECT
    A.ORGANIZATION_ID
    , C.CONVERSATION_XID
    , A.ANSWER_XID
    , I.QUERY_EXPRESSION
    , A.IS_QUERY_ANSWERED
    , A.ANSWER
FROM COVEO_CORE_MODEL_V001.COMMON.INSIGHTS I
JOIN COVEO_CORE_MODEL_V001.ANSWERING.ANSWERS A
    ON I.ORGANIZATION_ID = A.ORGANIZATION_ID
    AND I.INSIGHT_ID = A.ANSWER_ID
    AND I.INSIGHT_DATE = A.ANSWER_DATE
JOIN COVEO_CORE_MODEL_V001.ANSWERING.CONVERSATIONS C
    ON A.ORGANIZATION_ID = C.ORGANIZATION_ID
    AND A.CONVERSATION_ID = C.CONVERSATION_ID
WHERE A.ANSWER_SOURCE = 'search_agent'
AND I.INSIGHT_DATE BETWEEN '<START_DATE>' AND '<END_DATE>'
AND C.CONVERSATION_XID = '<CONVERSATION_XID>'

Where you replace:

  • <START_DATE> and <END_DATE> with date values, formatted as YYYY-MM-DD. A conversation can remain active for up to 48 hours, so use a date range that covers the full period of interest.

  • <CONVERSATION_XID> with the external conversation identifier from the Agent API (for example, conv_00991f9b-5fc5-43a3-a516-13fff1ebdfa0).