LLMs.txt directory

TS: Object Tracking system Model

Beta

The TS: Object Tracking system Model, sometimes referred to as Object History, gives a complete metadata view of every Liveboard, Answer, and Model that has issued a query: who created it, when, who last changed it, and how it is being used. It combines object metadata (author, created, modified) with the query telemetry behind TS: AI and BI Stats, so you can go from "this object is unused" to "here is who owns it" in a single search.

The Model has 53 columns: 9 measures and 44 attributes. It is available from release 26.3.0.cl.

Use Object Type to pick Liveboards, Answers, or tables. Use Object Modified Time and Object Author Name for ownership and staleness. Use User Action Time with Impressions or Active Users to see whether anything is still being used. Filter Is System = False to exclude background work.

Columns that start with Object describe the Liveboard or Answer itself. Columns that start with Model describe the Model it is built on.

Column reference

Column name Data type Category Description

Actions

int64

Measure

Number of user actions recorded. One Liveboard open with six visualizations records six actions, because the count is query-based.

Active Users

int64

Measure

Unique number of users who ran at least one query in the period (unique count of User ID).

Answer Session ID

varchar

Attribute

Internal ThoughtSpot session ID for a single Answer or Liveboard viewing session.

Connection

varchar

Attribute

Name of the live-query connection to your cloud data warehouse.

Connection ID

varchar

Attribute

Unique ThoughtSpot ID of that connection.

Credits

int64

Measure

Overall ThoughtSpot credit usage attributed to the query.

DB Auth Type

varchar

Attribute

Authentication method (for example, OAuth) used on the warehouse connection.

DB End Time

date_time

Attribute

When the external data warehouse finished processing the query.

DB Start Time

date_time

Attribute

When the external data warehouse started processing the query.

DB Type

varchar

Attribute

Type of data warehouse the query ran against (Snowflake, Databricks, BigQuery, and so on).

Error Message

varchar

Attribute

Error text returned when a query fails. Empty for successful queries.

Impressions

int64

Measure

Interactions with an object (unique count of Answer sessions).

Is Billable

bool

Attribute

True if the query counts toward usage and billing.

Is System

bool

Attribute

True if ThoughtSpot ran the query itself as background work. Filter to False to see genuine user activity.

Model

varchar

Attribute

Name of the Model the object is built on.

Model Author ID

varchar

Attribute

User ID of the person who created the Model.

Model Author Name

varchar

Attribute

Display name of the person who created the Model.

Model Created Time

date_time

Attribute

When the Model was created.

Model ID

varchar

Attribute

Unique ID of the Model.

Model Modified By ID

varchar

Attribute

User ID of the person who last modified the Model.

Model Modified By Name

varchar

Attribute

Display name of the person who last modified the Model.

Model Modified Time

date_time

Attribute

When the Model was last modified.

Object Author ID

varchar

Attribute

User ID of the person who created the object.

Object Author Name

varchar

Attribute

Display name of the person who created the object.

Object Created Time

date_time

Attribute

When the object was created.

Object ID

varchar

Attribute

Unique ID of the object. Use this to join to REST API output or trace an object across environments.

Object Modified By ID

varchar

Attribute

User ID of the person who last modified the object.

Object Modified By Name

varchar

Attribute

Display name of the person who last modified the object.

Object Modified Time

date_time

Attribute

When the object was last modified.

Object Name

varchar

Attribute

Name of the object as shown in the ThoughtSpot UI.

Object Subtype

varchar

Attribute

More specific type within an object type, for example, the individual Answers inside a Liveboard.

Object Type

varchar

Attribute

Type of object: LIVEBOARD, ANSWER, or table.

Org

varchar

Attribute

Name of the Org the object and activity belong to.

Org ID

varchar

Attribute

Unique ID of the Org.

Queries

int64

Measure

Number of queries run (unique count of ThoughtSpot Query IDs).

Query Errors

int64

Measure

Number of queries that ended in error or timeout.

Query ID

varchar

Attribute

Query ID assigned by the external data warehouse. Useful for tracing on the warehouse side.

Query Latency (External)

int64

Measure

Seconds the data warehouse spent on the query (DB End Time − DB Start Time).

Query Rows Fetched

int64

Measure

Number of rows the query returned.

Query SQL

varchar

Attribute

SQL statement ThoughtSpot sent to the data warehouse.

Query Status

varchar

Attribute

Whether the query succeeded or failed.

System Latency (Overall)

int64

Measure

Total seconds to answer the query end to end, including ThoughtSpot processing.

ThoughtSpot Query ID

varchar

Attribute

ThoughtSpot’s own unique ID for the query. Quote this when raising a support case.

Trace ID

varchar

Attribute

Follows one request across every ThoughtSpot layer. Diagnostic aid.

TS Query End Time

date_time

Attribute

When ThoughtSpot finished handling the query.

TS Query Start Time

date_time

Attribute

When ThoughtSpot began handling the query.

User

varchar

Attribute

Username of the person who ran the query or action.

User Action

varchar

Attribute

What the user did, for example, viewed a Liveboard, opened an Answer, ran a search, or had a Spotter conversation. See User action code reference.

User Action Time

date_time

Attribute

When the activity happened. Main date field for usage trends.

User Display Name

varchar

Attribute

Full name of the user as shown in ThoughtSpot.

User ID

varchar

Attribute

Unique ID of the user.

Visualization ID

varchar

Attribute

Unique ID of the visualization.

Vizualization Name

varchar

Attribute

Name of the chart or table the user interacted with.

The column name Vizualization Name contains a spelling error as it appears in the ThoughtSpot product. The spelling is preserved here to match the UI exactly.

Example searches

Search Returns

object name object type = 'LIVEBOARD' object author name object created time object modified by name object modified time org

Ownership register of every Liveboard per Org.

object name object type object author name max user action time org user action time > 180 days ago

Objects with any usage in the last 180 days. Everything not on this list is a candidate for archiving.

object name object type object modified time object modified time < 180 days ago org

Objects not modified in the last 180 days. The staleness half of an archiving rule.

model model author name model modified by name model modified time object name object type

Which Liveboards and Answers depend on each Model. Use this for impact analysis before changing a Model.

object author name unique count object id object type

Content footprint per author. Find owners with large estates before they leave or change role.

object name org impressions active users user action time monthly object type = 'LIVEBOARD'

Monthly usage per Liveboard per Org. An adoption view for a product team.

Use cases

Inactive content identification for migration and archiving

If you apply an inactivity rule, such as 180 days, this Model gives both halves of it: last modification (metadata) and last use (telemetry), per Org, with the owner to contact.

Reconciling "last viewed" discrepancies

Check max user action time for the object. Remember the two lags: metadata refreshes every 2–3 days, and telemetry is D-1. Names for Liveboards captured before February 26, 2026 may be null.

Change audit

Object Modified By Name and Object Modified Time answer "who changed this Liveboard last week" without opening each object.

Model impact analysis

Before altering a Model, list every Liveboard and Answer built on it and notify those owners.

Things to be aware of

  • The Model is in Beta. Column names are stable, but additional columns may be added.

  • Only objects that have issued at least one query appear. A Liveboard that was created but never opened is not listed until it is used.

  • Object Name is only populated for Liveboards for activity captured after February 26, 2026. Earlier rows show null.

  • The author and modified metadata refreshes every 2–3 days. The usage columns follow the daily query-stats refresh.

  • On large clusters, open-ended searches with no date filter can be slow. Always add a date range on User Action Time or Object Modified Time.