Live queries read a database when a question runs. Refreshed snapshots analyze the latest successful import. Uploaded files remain at the supplied version until replaced. A connected workbook can contain all three, so “connected” alone does not mean real time.
Three ways to bring data to AI
TableArth supports selected database queries, connectors with refreshed snapshots, and file or collection imports. Choose based on the source, required freshness, permissions, and reporting purpose.
| Source | Analysis uses | How to update |
|---|---|---|
| Supported databases | Authorized data queried on demand | Run a new query; results reflect that query’s execution time |
| Google Sheets | Snapshot of selected tabs | Refresh the connection and confirm success |
| Cloud billing and advertising | Imported provider reports or billing exports | Refresh the connector; provider exports can have their own reporting delay |
| Excel, CSV, and Tally imports | Data supplied in the file or collection export | Provide an updated file or run another import |
What a one-time upload gives you
An uploaded file is useful when you want a stable dataset for a review, reconciliation, or historical comparison. Record the file version, export date, time zone, and reporting period so another person can reproduce the analysis.
If the original file changes, an earlier upload does not change with it. Upload an updated version for a new analysis. A chart based on an older upload can still be correct for that version; label the period clearly.
What connected data changes
A configured database connection can query selected tables or collections on demand. This avoids exporting a file for each question, but query time, read permissions, and database workload still matter. A displayed answer remains the result of the query that produced it.
Google Sheets, cloud costs, and advertising reporting use snapshots in TableArth. Their freshness depends on successful refreshes and the upstream source. For example, a cloud billing export may arrive after the usage occurred. A successful import does not remove that provider delay.
Workbooks can organize multiple sources together. Cross-source analysis needs compatible query paths and matching relationships. Compare periods and definitions before joining a recent database result to an older spreadsheet snapshot.
When uploads are the right choice
- A fixed month-end export that should remain reproducible.
- A one-off analysis of a CSV or Excel workbook.
- A Tally collection export for a defined voucher date range.
- A dataset that cannot be connected directly under your access policy.
Choose a refresh routine
- Identify whether the source is queried live, refreshed as a snapshot, or imported manually.
- Agree on the freshness needed for the business question.
- Check the successful refresh or import and the source reporting period.
- Validate totals, units, time zones, and matching keys before combining sources.
- Label the period and source when sharing a dashboard.
Explore the connector directory, learn how workbooks organize analysis, or compare database queries with Google Sheets snapshots.
Is uploading a file bad?
No. Uploading is the right choice for a genuine one-off — a quick check before a meeting, or trying tableArth.ai for the first time on a sample file. It becomes the wrong choice only when you keep doing it for the same source, because each upload is a fresh snapshot someone has to keep replacing by hand.
How does live sync work?
Databases can be queried on demand. Google Sheets, cloud billing, and advertising use refreshed snapshots. Uploaded Excel, CSV, and Tally data reflects the imported version until updated.
Do I ever need to re-upload?
Yes, when you need newer data from a file upload or Tally collection export. A configured snapshot connector can refresh selected data; check that it completed successfully.
What about large files?
Large uploads are handled reliably. The timeout issues that used to affect bigger files are fixed, so the upload process is smoother and more consistent for large datasets — and for anything that keeps growing, connecting the source instead of re-uploading it is still the more reliable long-term path.