The rules a filter follows
The example keeps paid orders above 100. Order 1008 has no amount: it is missing, not zero, so it fails “> 100” and is listed in the ledger instead of slipping through or causing an error.
A missing value never satisfies a comparison, not even “does not equal”. Use “is missing”, “is empty text” or “is blank” to target gaps deliberately. Numeric conditions use exact decimals, so 100.00 equals 100 and a 19-digit ID compares correctly.
Exports always state their scope, for example “final output after step 1: 5 rows”, and never export just the rows visible in the grid.
Questions
Can I filter by a regular expression?
Yes, with “matches regex”. The pattern runs in the worker with a time budget and the step editor has a test bench for trying it on sample values first.
How do I filter by date?
Date conditions expect ISO dates (YYYY-MM-DD). If your data uses another format, add a “Convert type / date” step first with the format stated.
Can I see what was filtered out?
The Ledger tab lists removed rows for the selected step, and “Export removed rows” saves them as a CSV.