What a clean step does
A clean step is the general-purpose step: filter rows, rename fields, remove fields, change types, split values, group values and add calculations. Most flows have several.
Everything you do is recorded in the Changes panel on the left, in order, and each entry can be edited or removed. That list is the documentation of what the flow does — and unlike a SQL script, it cannot get out of step with what actually runs.
Profile-pane actions
The fastest cleaning happens by right-clicking values in the profile pane:
- Exclude — filter out that value
- Keep only — filter to it
- Edit value — correct a single wrong value in place
- Group — merge several values into one
- Replace with null
Editing a value directly is the quickest fix for a handful of typos, and it records itself as a change rather than being a hidden manual edit.
Filtering
Beyond value-level filters, the toolbar offers range filters, wildcard match, null filters and a calculated filter for anything more complex.
Every row removed early is a row every later step does not process. Filters belong as close to the input as the logic allows.
Field cleaning options
The ... menu on a field header offers Clean actions — remove letters, remove numbers, remove punctuation, trim spaces, make uppercase or lowercase.
Trim spaces and set consistent case on any field you will join on. Trailing whitespace is the commonest reason a join finds no matches, and it is invisible in the data grid.
Keeping the flow readable
Give each step a name describing what it does — Clean customer names, Filter to active orders. The default Clean 1, Clean 2 tells a reader nothing, and a flow of fifteen steps quickly becomes opaque.