130 screens carry a CSV import. It is the same four steps everywhere, and nothing is written until the last one.
Note
Not every screen has it. If the ⬆ Import button is not on the toolbar, that screen is keyed by hand.
Step 1 — Upload
- Open the screen you want to load and press ⬆ Import.
- The page is headed "Import <screen name>" with the four steps shown across the top: 1 · Upload → 2 · Map columns → 3 · Check → 4 · Import.
- Choose your CSV file. The first row must hold the column headings.
- Press Upload & continue.
Note
Up to 5,000 data rows per file. If an earlier attempt is still pending the page offers to continue with it or replace it.
Step 2 — Map columns
The mapping table lists every column in your file with three things: the CSV column heading, the first row's value as an example, and an "Imports into" selector. The importer has already guessed the mapping from your headings; correct anything it got wrong, and set the columns you do not want to "— don't import —".
Below the table
- A "Not mapped" note listing the columns that will be ignored.
- On screens with a unique code, a choice of what to do when a row's code already exists: Skip it (existing records untouched, only new codes added) or Update it (the mapped columns overwrite the existing record). Update needs the edit permission on that screen.
- Press Check the rows.
Step 3 — Check
This is a dry run — the sub-heading says so: "dry run of N rows — nothing has been imported yet". You get counters for Will be added, Will be updated or Duplicates skipped, and Rows with errors, plus a table of the rows that would be left out.
Step 4 — Import
- Press "Import N rows".
- Confirm the prompt.
- The valid rows are written in a single transaction; the invalid ones are left out.
Note
Every run is recorded — the import log and the workspace activity trail both keep it, so you can see afterwards who loaded what and when.