Preparation node
Unique
Split rows into first-seen (Unique) and repeats (Duplicate) by key columns.
What it does
Splits the rows into first-time and repeat. The first time a key combination appears the row leaves the Unique anchor; every later row with that same key leaves the Duplicate anchor. Nothing is thrown away: the two outputs together are exactly the input, which is what lets you deduplicate and inspect what was removed in one node.
Inputs and outputs
| Inputs | One. |
| Outputs | Unique, Duplicate. |
When to use it
- Deduplicate a list and keep one row per customer, order or key
- Find the repeats: wire up the Duplicate anchor and look at them
- Prove a file has no duplicates before loading it somewhere strict
Examples
Dedupe on a key
Key: email
First row per email leaves Unique; every repeat leaves Duplicate.
Fully identical rows
Every column (the default)
Removes exact duplicate rows, the usual fix after stacking files.
Composite key
Keys: order_id + line_no
A row is a repeat only when EVERY key column matches.
Choose which one wins
Sort by updated_at Descending → Unique on customer_id
Unique keeps the first row it sees, so the sort decides the survivor.
Tips
- By default EVERY column is the key, which means exact duplicate rows. Untick columns to key on the ones that are left; the key needs at least one column.
- Text keys are compared case-insensitively unless you turn on Match case on text keys, so “Apple” and “apple” count as the same key.
- Unique keeps the FIRST occurrence in input order. If it matters which row survives, Sort first.
- The Duplicate anchor is optional: leave it unwired and the repeats are simply dropped.
- Deduplicating on one key column keeps the whole row, not just the key. The other columns come from whichever row won.