Create datasets from CSV and query snapshots

You may have CSVs, Parquet files, or other one-off data that needs to be queried alongside warehouse tables. You can save these files as datasets and access them through the same interface as connected sources. You can upload a file, derive a dataset from a query, inspect its structure, and join it with data from other systems.

How it works

Step 1 — Upload from a file

Upload targets.csv as a dataset titled Q2 targets with table name q2_targets—what do I type in SQL to read it?

Step 2 — Or materialize a query you already trust

Save a table built from billing data: everyone who’s still subscribed, with id, email, and plan. Title it Active subscribers and name the table active_subscribers.

Step 3 — Inspect before you join

Show my uploaded datasets, then open the one I pick so I can see each field and type.

Step 4 — Query like any other table

Match my Q2 targets upload to sales accounts on account id—give me account_id and plan for up to 50 pairs.

Who uses this

• Operations importing periodic CSV extracts without warehouse DDL changes.
• Analytics saving cohort definitions as stable dataset tables.
• Automation that queries uploaded tables by name rather than unparsed attachments.