Output types
| Type | Use for |
|---|---|
| .hyper extract | Local file for Tableau Desktop |
| Published data source | Shared on Server or Cloud — the usual choice |
| CSV / Excel | Handing data to other tools or people |
| Database table | Writing back to a warehouse |
Published data source
The most useful output for a team. The flow writes to Server or Cloud, and every workbook connecting to that data source gets the refreshed data.
That gives one prepared, governed data source rather than each analyst preparing their own — which is normally the reason to adopt Prep at all.
Write options
For file outputs: Create table replaces, Append to table adds rows.
Append adds every row again each time the flow runs. Unless the flow only ever processes new rows, use create/replace.
For database outputs there is also Replace data, which keeps the table structure and permissions while swapping the rows — usually better than dropping and recreating.
Several outputs
One flow can have several output steps on different branches: a full detail extract, an aggregated summary, and an exceptions file of rows that failed a join.
That last one is worth building as a habit — rows dropped by a join should go somewhere a person can see them rather than vanishing.
Field naming
The output is what analysts see in Tableau. Rename fields to what a business user would call them — Revenue, not rev_amt_usd — and remove anything nobody needs.
A prepared data source with 200 cryptic fields is barely better than the raw table.