Join tables and define relationships across datasets. Use Relations to look up matching values across tables, or Merge to combine tables that share the same columns.
When you have more than one table open in the transformation workspace, you can connect them two ways. Relations link two tables on a matching column, similar to a database join, so you can pull values from one table into another. Merge stacks two or more tables that share the same column names into a single output. Both are available from the toolbar once you have multiple tables in the Tables panel.
A relation links the active output to another table in your workspace by matching a column in each-for example, matching a "Name" column in both tables to pull in additional fields.
The join type controls which rows are kept once the two tables are related.
| Join Type | Result |
|---|---|
| Inner join | Keeps only rows where the matching column has a value in both tables. |
| Left join | Keeps every row from the current output, filling in matches from the related table where available. |
| Right join | Keeps every row from the related table, filling in matches from the current output where available. |
| Full outer join | Keeps every row from both tables, leaving unmatched fields empty where no match exists on either side. |
If rows are missing after creating a relation, check the join type first. Inner join drops unmatched rows on both sides, while Left, Right, and Full outer join each keep unmatched rows from at least one table.
Beyond exact matches, relations support fuzzy matching for columns where values are similar but not identical between tables-useful for names, SKUs, or free-text fields that vary slightly across sources.
| Match Type | Best For |
|---|---|
| Exact | Values must be identical, character for character. |
| Case insensitive | Values must match except for uppercase or lowercase differences. |
| Levenshtein | Allows values that differ by a set number of character edits, controlled by the tolerance value. |
| Jaro-Winkler | Allows values that are similar overall, weighted toward matching prefixes-useful for names and short text. |
The tolerance value next to the match type only applies to Levenshtein matching. Start with a low value and increase it gradually-a value that's too high can create false matches.
Merge combines two or more tables that share column names into a single output-useful for stacking files with the same structure, like monthly exports or files from different sources with matching headers.
Select 2 or more distinct tables to merge.
Merge matches columns by name, not position. Tables with differently named columns for the same data should be aligned first-for example by renaming columns in the Output columns panel-before merging.
If you need assistance creating relations or merging tables, our support team is here to help.
Contact Support