Column mapping is the first logical step in turning raw supplier data into usable product information. It translates the column names in a supplier's feed into the canonical attribute names defined in your workspace schema.
How It Works
Every supplier uses their own naming conventions. One supplier might call a field colour_name, another Color, and a third clr. In your schema, that attribute is simply color. Column mapping creates the translation layer between what the supplier sends and what your schema expects.
Mappings are stored as key-value pairs on the supplier's adapter: each pair maps a raw column name from the feed to a canonical attribute name in your schema.
| Raw Column (Supplier) | Canonical Attribute (Your Schema) |
|---|---|
| colour_name | color |
| Gewicht | weight |
| Produktname | name |
| Mat. | material |
| Breite | width |
Mapping Priority
When FeedPrep processes a feed, column names are resolved using the following priority order:
- Rule-based mappings (highest priority) — Explicit rules you have defined take precedence over everything else.
- Supplier adapter mappings — Mappings saved on the adapter for that specific supplier.
This means you can set broad adapter-level mappings for a supplier and override specific columns with rules when needed.
Unknown Columns
When a feed contains columns that don't match any existing mapping, FeedPrep doesn't silently discard them. Instead, it creates inbox items alerting you that unmapped columns were found. You can then review each unknown column and decide which canonical attribute it should map to — or ignore it if the column isn't relevant.
This ensures new or renamed columns from suppliers never slip through unnoticed.
One-to-One Mapping
Each canonical attribute can only be mapped once per adapter. You cannot map two different raw columns to the same canonical attribute. This prevents duplicate or conflicting data from entering your normalized output.
Automatic Reuse
Once you map a column on a supplier's adapter, that mapping applies automatically to every future feed from that supplier. You do the work once; FeedPrep remembers it forever. If Supplier A always calls their color column colour_name, you only need to map it the first time.
AI-Assisted Column Mapping
FeedPrep's AI can suggest column mappings for unmapped columns by analyzing the raw column names against your schema's attribute names. This is especially useful when onboarding a new supplier with many columns.
AI column mapping suggestions cost 2 credits per request. You always review and approve suggestions before they are applied — the AI never maps columns without your confirmation.