Data Last Updated
Data Last Updated shows when the source tables/views behind a Data Mart last changed in the storage. It answers the question “how current is what I am looking at?” before you build a report or act on an answer from an AI assistant.
Data Last Updated is currently measured for Google BigQuery Data Marts. Data Marts on other storages show Unknown until their support lands.
What the value means
Section titled “What the value means”Read the value precisely. It is a storage timestamp, not a statement about the data’s content:
- It says when something last wrote to the source tables. A backfill can rewrite a table today with figures from 2021, so “updated today” does not mean “covers today”.
- Unknown means OWOX could not determine the time. It does not mean the data is stale — or fresh.
- A value marked ≥ (“at least as recent as”) means some source tables could not be checked. The real time can only be more recent than shown.
Hover over any Data Last Updated value or icon to see the details: the exact timestamp, when OWOX ran the check, how complete the coverage is, and the per-table breakdown.
Where you see it
Section titled “Where you see it”| Surface | What is shown |
|---|---|
| Data Mart page → Overview → Details | A Data Last Updated tile with a Check now button |
| Data Marts list | A sortable Data Last Updated column |
| Models canvas | A clock icon on every node, next to the Data Quality shield, plus a Data Last Updated entry in the Actions menu |
| MCP | A data_last_updated block in every query_data_mart response — see MCP Server |
The list and the canvas show the last-known value — the result of the most recent check or data-delivering run, with its time visible in the hover card. They never query the storage on page load.
How to check it
Section titled “How to check it”Two manual checks measure the value live and save the result:
- Check now (⟳) on the Data Mart page measures that one Data Mart.
- Actions → Check Data Last Updated on the canvas measures every Data Mart currently visible on the canvas, in one pass.
Checking is free. It reads storage metadata only, consumes no credits, and records no run. Any project member who can see the Data Mart can run the check.
You rarely need the buttons, though: every run that delivers data refreshes the value on the way out — see the table below.
A check that cannot determine the time reports Unknown and keeps the previously saved value — a failed lookup never erases the last-known answer.
When the value updates
Section titled “When the value updates”| Trigger | Effect |
|---|---|
| Check now on the Data Mart page | Measures and saves the value |
| Actions → Check Data Last Updated on the canvas | Measures and saves the value for all visible Data Marts |
| Report run — manual or scheduled | Measures during the run and records the result in the run’s history entry. When the report reads only the Data Mart’s own sources, it also saves the value |
| HTTP Data request | Same as a report run: records in history, saves when the request reads only the Data Mart’s own sources |
MCP query_data_mart call | Measures live for that response and records it in the run’s history entry. Saves the value when the query reads only the Data Mart’s own sources (no joined fields) |
| In the background, on its own | Never. OWOX only measures when someone checks manually or a run delivers data — a scheduled report run counts as a run |
A blended run records the measurement in its history entry only. Its reading covers other Data Marts’ tables too, so saving it on one Data Mart would overstate that value.
How OWOX determines the value
Section titled “How OWOX determines the value”OWOX asks the storage which tables the Data Mart reads, then takes the newest modification time among them:
- Views and SQL Data Marts resolve through to their underlying base tables, nested views included. Views themselves are excluded from the per-table breakdown — a view’s own modification time reflects a change to its definition, not to any data.
- Sharded and wildcard table sets (
events_YYYYMMDD) collapse into a single entry showing the newest shard. Neighbouring tables that merely share the prefix, such asevents_backup, do not count. - External tables (for example, Google Sheets) have no modification time in the storage. They appear in the breakdown as unknown, and the coverage becomes partial.
- Queries referencing more than ~50 tables are reported with partial coverage, because the storage truncates the list.
Data Last Updated vs. the Data freshness check
Section titled “Data Last Updated vs. the Data freshness check”These are different tools that answer different questions:
- Data Last Updated is storage metadata: when the source tables were last written to. It requires no configuration and costs nothing.
- The Data freshness check in Data Quality Checks validates the data itself: it reads a timestamp field and alerts when its latest value is older than a threshold you configure.
A table can pass one and fail the other. A backfill makes Data Last Updated show “today” while the Data freshness check still fails, because the newest timestamp inside the data is old.