Column mapping
Mapping your columns to system fields, which fields exist, and what happens to the columns you don't map.
Your columns rarely share names with the fields MarketplaceHub uses, so an import includes a mapping step. Matches are proposed for you to confirm or change — nothing is renamed in your file.
The system fields
| Field | Required | Notes |
|---|---|---|
| SKU | Yes | The join key for the entire system. |
| Title | No* | Not enforced at import, but no marketplace will accept a listing without one. |
| Description | No | |
| Price | No | |
| Quantity | No | |
| Cost | No | Unit cost. Used only as the margin floor when automated repricing is on. |
| Barcode | No | UPC, EAN or ISBN. Type is inferred from the value. |
| Category | No | Your own category, not the marketplace's. |
| Main image | No | Primary gallery image, as a URL. |
| Image 1–24 | No | Additional gallery slots, as URLs. |
What happens to your other columns
They are not discarded. Unmapped columns are carried through with the product and offered to the AI when it fills in whatever the destination marketplace's product type requires — so a wide export helps rather than hinders.
This is why the mapping step is short even for a 90-column feed: you are mapping the handful of fields MarketplaceHub itself needs, not every field the marketplace will eventually want. Those come from the field reference, and are filled by classification and review.
Changing a mapping later
Mapping belongs to the import, not to one upload, so it applies on every subsequent run. Changing a step invalidates the steps after it, which are then re-run — the alternative would be a configuration whose later stages silently describe data that no longer exists.