CSV and file uploads Integration Guide

Before you start

Permission to upload/import data in the intended Data Weaver client and a file you are authorized to process.
Prepare readable headers and review dates, units and currencies. Remove unnecessary personal or confidential information before uploading.

Select the correct client in Sources. For a standard CSV metric import, choose Upload CSV, select the data type and explicitly map file columns to the supported fields before confirming.
For arbitrary tabular evidence, use Uploaded datasets: upload a CSV file, review its rows/columns and interpretation, name the dataset and confirm the import.
Keep arbitrary datasets separate from standard marketing totals unless their meaning has been explicitly reviewed. Select the dataset when asking a dataset-specific question.

No OAuth scopes or provider account permissions are used. Uploading authorizes Data Weaver to process the supplied file in the selected client context.
You are responsible for the right to use the file. The public guide does not read your files or call authenticated APIs.

Standard mapped CSV imports populate supported metric structures. Flexible datasets retain tabular evidence and reviewed interpretation separately for scoped analysis.
A file upload does not authorize ongoing access to the original platform or make inferred business meanings definitive.

There is no automatic provider refresh for local files. Upload a new file or use the supported re-map/re-upload flow and review the mapping again.
For spreadsheets read directly from Google, use the separate Google Sheets guide and reviewed Refresh workflow.

Both local upload paths accept CSV. Export workbooks to CSV first; renaming a workbook does not convert it. Flexible datasets are limited to 8 MiB, 50,000 data rows, 100 columns and 1,000,000 cells, and the encoded import must also fit the request limit. Follow the separate validation messages for the standard metric-import path.
Arbitrary dates, duplicate rows, summary rows and currencies need review. File evidence should not silently blend with connected-provider totals or imply complete history.

If parsing fails, export a clean supported tabular file with consistent headers and inspect validation messages. A standard CSV upload expects CSV, not a renamed workbook.
If totals look wrong, review column mapping, date/currency interpretation and exclusions. Re-uploading is not OAuth reauthorization.

There is no provider grant to revoke. Delete the upload or dataset in Data Weaver to remove its imported data through the relevant flow.
Deleting imported data does not automatically remove past answers or exported copies. Use Data deletion for wider retained application data; manage the original source file separately.

You stay in control of what you connect. Review how we handle information, request deletion, or contact us.