⚠️ An update replaces the whole record. The new record is built from that CSV row alone, so a
column you did not map — or left empty — clears the value it had before, and a row with no link
columns removes the record's existing links. Always import the complete row, not just the
columns that changed. Correcting one column means re-sending the whole row with that column fixed.
| Rule | Limit |
|---|---|
| Encoding | UTF-8 (a byte-order mark is accepted) |
| Separator | Comma or semicolon |
| File size | 10 MiB |
| Data rows | 10,000, not counting the header |
| Columns | 256 |
"") and line breaks.product, brand, category or collection, or Link → other type… for a| Field type | Accepted values |
|---|---|
| Text | Imported as written, including leading zeros. |
| Number | . as the decimal point, for example 12.5. Scientific notation (1e3) is accepted. Values too large to store exactly are rejected. |
| Yes / No | true / false (any case), or 1 / 0. |
| Date & time | YYYY-MM-DD, or a date and time with seconds (up to 3 decimals allowed) and Z or an offset, for example 2026-02-28T23:00:00+02:00. Converted to UTC; a date on its own means midnight UTC. |
|, for example id-1|id-2.| Status | Meaning |
|---|---|
| Ready | Valid, not sent yet. |
| Invalid | Failed validation and was not sent. The row shows why. |
| Created | The record was created. |
| Updated | The row's key matched an existing record, which was replaced with the new values. Only on a table with a key field. |
| Already imported | A retry found the record already created. Only on a table without a key field. |
| Rejected | The platform refused the row, for example because it exceeds the 64 KB record limit. |
| Unconfirmed | The outcome is unknown, typically after a network error. |
codesku, before importing anything. Then every later import updates the right records instead of