The join step
The join summary — the best feature in Prep
The join step displays a Venn diagram with counts: how many values matched, and how many are unique to each side. Click the unmatched portion and the profile pane shows exactly which values failed.
In SQL an inner join silently discards non-matches. Here you see the count immediately, and one click tells you they are all trailing-whitespace variants — which is a genuinely better workflow for data preparation.
Watch the row count
The step shows the resulting row count. If it is higher than the left input, the right side has duplicate keys and rows have fanned out — every measure downstream is now inflated.
Fix it by adding an aggregate step on the right branch first, grouping by the join key so there is one row per key.
Union
Drag one step onto another and choose Union. Prep matches fields by name and shows a field mismatch panel listing anything present in some inputs but not others.
Mismatched fields can be merged manually where the same thing has different names — cust_id in one file and customer_id in another.
Prep adds Table Names and File Paths fields so you can tell which input each row came from. Keep them until you have verified the union, then remove them if not needed.
Which to use
Union stacks rows from similarly shaped sources. Join matches rows across sources on a key. If you are unioning things with different fields, that is usually a sign one of them should have been joined instead.