The Check step validates every row against the same rules the screen's own form uses. Nothing is different about an imported record — a row that the form would refuse, the importer refuses too.
Blocking problems
Some problems stop the whole file rather than a row. They appear in a red panel headed "This file cannot be imported yet", with the hint to go back to mapping and assign the missing field.
- A required field of the screen has not been mapped to any column. The message names the field: "The required field \"Party Code\" is not mapped to any CSV column."
- The file could not be read at all — empty, no header row, a header row with no data under it, or an encoding the reader does not recognise.
- More than 5,000 data rows. Split the file and import the parts one at a time.
Row errors
Everything else is reported per row, in a table headed "Rows that will be left out" with three columns.
| Column | What it tells you |
|---|---|
| Row | The line number in your file — the header is line 1, so the first data row is 2. |
| Column | Which of your CSV columns the problem is in. |
| Problem | The validation message, in the same words the screen's own form would use. |
Note
Up to 200 errors are listed; beyond that the panel says how many more there are. Long lists usually mean one systematic problem, not 200 different ones — fix the pattern and re-check.
Typical causes
- A code that points at a record you have not imported yet — the classic symptom of importing out of order.
- A date the reader cannot parse. Use YYYY-MM-DD.
- A value longer than the field allows, or a number with a thousands separator in it.
- A blank in a required column.
Clearing them
You have two ways forward, and the panel offers both: fix the file and re-upload, or import the valid rows now and deal with the rest afterwards. When only a handful fail, importing the good rows and keying the remainder is usually quickest.