How to map CSV columns to product catalog fields
Every supplier names their columns differently — Item#, sku_code, PartNumber — and mapping them by hand is where catalog projects stall. This guide walks the full workflow in InventaCloud: reviewing your source file, uploading it, checking the AI's mapping suggestions, applying validation rules, and importing a clean catalog you can reuse on every future file from that supplier.
1. Review the source CSV or Excel file
Before uploading anything, open the file and get oriented. You are looking for three things: which row holds the headers, whether every product has a unique SKU, and which columns actually matter for your catalog. Supplier files often carry dozens of internal columns you will never use — that's fine; you only map what you need.
Both CSV and XLS/XLSX files work. If your supplier sends something else, ask for one of those two formats — they're the file types the import workflow accepts today. Prefer a video? The upload and mapping video tutorials walk this same workflow on screen. This guide goes deep on the mapping stage — deciding what each supplier column becomes; for the complete import lifecycle around it (file preparation, validation, import progress, and post-import review), see how to import a vendor catalog from CSV or Excel.
2. Identify the important source columns
The columns most catalogs map, and what to watch for in each:
| Catalog field | Typical source column names | Watch for |
|---|---|---|
| SKU | SKU, Item#, PartNumber, sku_code | Must be unique per product — it's the matching key |
| Product title | Name, Description1, ItemName | All-caps or truncated names (cleanup rules can help) |
| Description | Description, LongDesc, desc2 | HTML fragments or line breaks in the text |
| Brand | Brand, Mfr, Vendor | Inconsistent spellings of the same brand |
| Category / product type | Category, Dept, Class | Supplier categories rarely match yours one-to-one |
| Cost | Cost, DealerPrice, WholesalePrice | Confusing cost with selling price |
| Price | Price, SellPrice, Retail | Currency symbols and thousands separators |
| MSRP | MSRP, ListPrice, SRP | Blank cells where no MSRP exists |
| MAP | MAP, MinAdvertised | Only map this if the supplier actually provides it |
| Inventory | QtyOnHand, Stock, Avail | Text values like "in stock" instead of numbers |
| Weight | Weight, ShipWeight | Mixed units (lb vs kg) across rows |
| Dimensions | L / W / H, Size, Dims | Three columns vs one combined "10x4x2" column |
| Image reference | Image, ImageFile, Photo | Catalog imports take image filenames that match files in your media library — see the note in step 7 |
| Video reference | Video, VideoFile, Video1 | Video filenames (MP4, MOV, WebM) that match files in your media library; Video 01 to Video 10 are import-rule roles, so a product can carry several |
| Document reference | SpecSheet, Instructions, Manual | PDF filenames that match files in your media library; the roles are spec sheet, instructions, install guide and warranty, plus up to five general documents |
| Product status | Status, Active, Discontinued | Codes like "D" or "0" that need translating |
3. Upload the file
In the app, upload the CSV or Excel file from the catalog files area. Large files are queued and processed in the background with visible progress — you don't need to keep the page open. A newer version of the same file can be queued as a scheduled replacement to swap in at a set time.
4. Assign the vendor and catalog
Tell InventaCloud which vendor the file belongs to and which catalog(s) it should feed. This matters beyond bookkeeping: the saved mapping you're about to create belongs to this source, so the next file from the same vendor can reuse it. One source file can power multiple published catalogs.
5. Review the AI mapping suggestions
InventaCloud reads the headers and sample values and proposes a mapping from your file's columns to catalog fields. A column called "desc2" containing "blue widget 9.99" still gets a sensible suggestion because the values are analyzed, not just the header names.
Treat suggestions as a strong first draft, not a verdict. Common, clearly labeled columns are often suggested correctly on the first pass, but ambiguous ones — two price-like columns, say — need your judgment. Nothing publishes until you approve the mapping.
6. Correct and lock the mappings
Fix anything the AI got wrong: reassign a column, leave irrelevant columns unmapped, and confirm the SKU column above all — it's the key that keeps re-imports updating products instead of duplicating them. Once approved, the mapping is locked to this source and pre-filled on future files for a quick confirmation, so the heavy review is a one-time cost per supplier.
7. Apply import and validation rules
Rules are where supplier quirks get fixed permanently: set defaults for blank fields, replace or translate values (that "D" status code becomes "Discontinued"), normalize units, and exclude rows you never want imported. Field standards validate each row so wrong-but-well-formed data gets flagged instead of imported.
A note on images, videos and PDFs: catalog imports reference media by filename, matched against your media library (images, product videos and PDF spec sheets or instructions share one library) — bulk-upload the photos there (folders and ZIPs upload in one go, or give your reps a direct upload inbox) and the filenames in your CSV connect them. Arbitrary image URLs inside the catalog file are not imported.
8. Preview the mapped data
Before importing, look at the mapped result — real products with your field names and cleaned values. This is the moment to catch a swapped cost/price column or a category that mapped oddly, while it's still a two-minute fix.
9. Start the import
Run the import. Processing is queued with progress visible; valid rows import while problem rows are collected for the error report. Change detection works out what's new, changed, or discontinued compared to the previous import of this source.
10. Review results and errors
The result report tells you exactly what happened — rows imported, rows skipped, and per-row errors with the field that caused them. Fix issues at the source (or with a rule, so the fix is permanent) and re-import; SKU keying means re-running is safe.
11. Save and reuse the mapping
Your approved mapping, rules, and validations now belong to this source. The next file from the same supplier arrives with the saved mapping pre-filled for you to confirm — and if the supplier changes their layout, changed columns are detected and re-proposed for review while known fields keep their mapping.
Mapping checklist
- Headers identified and SKU column confirmed unique
- Cost vs price vs MSRP assigned to the right roles
- Irrelevant columns deliberately left unmapped
- Status codes translated by a rule, not by hand
- Units normalized (weight, dimensions)
- Image, video and PDF filenames match files in your media library
- Preview checked before the first import
- Error report reviewed after it
Common mapping mistakes
The most expensive mistake available. Check sample values: if the "price" looks too low, it's probably cost. Price roles keep them separate once mapped correctly.
Supplier headers lie — "Description2" might hold the real product name. The AI reads values for exactly this reason; when reviewing, you should too.
If the supplier's "SKU" changes between files, re-imports can't match products. Pick the truly stable identifier — sometimes that's the manufacturer part number.
Hand-edits vanish on the next import. A replace/normalize rule fixes the same problem on every future file — permanently.
Common questions
Does every column map automatically?
No — and you wouldn't want it to. The AI proposes a mapping for the columns it recognizes — common, clearly labeled columns are often suggested correctly — and you review, correct, and approve before anything is used. Ambiguous columns are exactly where your review earns its keep.
What file formats can I import this way?
CSV and XLS/XLSX — uploaded directly from your computer. If a supplier sends another format, converting to CSV first is the reliable path.
What happens when my supplier changes their column layout?
Changed columns are detected on the next import and re-proposed for your review; unchanged fields keep their saved mapping, and your rules carry over. You review the delta, not the whole file.
Can I undo an import that went wrong?
Price, inventory, and discontinued-status changes keep history with supported rollback, so a bad update to those fields can be restored. The better habit is the preview step — most wrong imports are visible there before they happen.
Map your first file in minutes.
Start a 14-day free trial and bring your messiest supplier CSV — onboarding call included.