Start with the meaning of each column
Review imported column types so identifiers remain text and quantitative values can be analyzed as numbers. Filter rows, inspect descriptive statistics and compare categories. A currency label does not convert values or supply an exchange rate.
For example, keep a customer code such as 001 as text. Converting it to the number 1 would lose information that matters when matching another dataset.
Clean and join with a review step
Review proposed cleaning changes before applying them. Interpolated values remain identified as estimates, with cleaning history available. Join two workspace datasets using exact-text left or inner keys, select right-side columns and inspect output counts.
Repeated keys can expand a join and inflate a total if you treat every resulting row as a new sale. The preview shows matched, unmatched and expanded rows so you can examine that effect before creating the independent result.
Keep the analysis connected to its source
Create a Model workbook or Compose report from the current analysis. Joined-source revisions and inherited estimate warnings travel with the supported snapshots. Subsequent edits to a joined dataset are independent; rerun and review the join to include later source changes.
Start with a clear workflow
- Import a CSV and review headers, identifiers and number columns.
- Inspect missing values and review any cleaning changes.
- For a join, choose two datasets, exact keys and the right columns to include.
- Review row counts and warnings, then create a new dataset, workbook or report.
Know what to expect
- CSV imports support up to 1 MB, 2,000 data rows and 26 columns, subject to document limits.
- Joins are limited to 2,000 output rows and 26 combined columns. Blank keys do not match.
- Descriptive statistics and interpolated estimates do not establish causes or forecasts.
- A saved join is a reviewed snapshot, not a live database connection or automatic external refresh.
Common questions
Will 001 match 1 when I join two datasets?
No. Join keys use exact text, including case and spaces. This preserves identifiers. Clean or standardize the source values deliberately if you intend them to match.
Does a join change either source dataset?
No. It creates a new independent Studio dataset and records both source revisions. Later source changes require a new reviewed join.
What happens when a key appears twice on the right?
Each matching right row produces an output row. The preview reports duplicate expansion so you can review its effect before interpreting counts or totals.

