Data Connection and Extract

A data connection is the link between a BI tool and the data it reads. It can stay live, querying the source each time a view is drawn, or it can be captured as an extract — in Tableau’s words, “a subset of information that is saved separately from the original dataset”.

Tableau gives two reasons to take an extract: “to enhance performance and to utilize Tableau features that may not be available or supported in the original data”. The speed argument is measurable — “Interacting with views that utilize extract data sources results in better performance compared to views connected directly to the original data” — because the tool reads a local optimized copy instead of asking a remote system every time.

What you trade is freshness, and this is where the concept earns its place in a glossary rather than a manual. An extract is a snapshot of a moment. It does not know the source has changed, and it will keep serving the old numbers with no visible sign that they are old. A dashboard built on a stale extract is not broken; it is confidently reporting last month.

The choice follows the question. File-based analysis of a fixed historical dataset loses nothing by extracting. Anything monitored — today’s orders, current error rates — needs either a live connection or a refresh schedule someone actually owns.

One consequence is easy to miss when sharing: an extract can contain underlying rows that are not visible in the displayed aggregates. Saving a workbook to Tableau Public creates an extract, and the visualization and its underlying data are downloadable by default. Download options can be changed, but Tableau Public is intended for public data. Other Tableau publishing environments have their own permissions; check the actual sharing path before publishing.

References:

• Extract your data