Troubleshoot

Import failed

Rows rejected, nothing matched, or duplicates created: the causes in order.

The import of 8,000 rows has finished: 5,960 created, 0 updated, 2,040 failed. The tempting move is to start on the failed rows. Do not. Your catalog already held 6,000 of these products, so zero updated means the SKUs did not match, and the 5,960 "created" are duplicates.

The result screen and its failure reasons are explained in Import products in bulk. This is the diagnosis order when the numbers surprise you, most frequent cause first.

1. Nothing matched, or duplicates were created

You re-imported products that exist, and updated is 0 while created is high. The import keys on sku, and the column did not match: Excel ate the leading zeros (0041711 became 41711), a prefix, or trailing whitespace.

Check: in PIM › Products, type a SKU from your file into Search SKU or name. If the product turns up under a different spelling, this is the cause.

Fix: correct the SKU column in the source (format it as text first) and re-run. Matched SKUs update rather than duplicate.

A day later, the duplicates sit at the top of the grid, which sorts by Updated, newest first. Delete them before anybody files or prices them (a model with variants cannot be deleted until its variants are gone), then fix the SKU column and re-run.

2. Rows were rejected with a reason

The first two reasons account for most rejected rows.

ReasonCheckFix
Unknown attributeHeader is a label (Gewindegröße), not the code (thread_size)Rename to the attribute code
Invalid optionSelect column holds the label Stainless steel V4A, not the code v4aUse the option code, or add the option first
Unknown familyfamily code misspelled or missing; required when creatingCorrect the code
Validation failedValue breaks a min/max or patternFix the value, or relax the validation
Unknown recordA reference names a record that does not exist yetImport the records first
Missing requiredA required attribute is emptyFill it, or drop the requirement

Fix the source file, not the exported error file, and re-run.

3. The values landed, and they are wrong

The result reports success and the data is damaged: Größe arrived as Größe (not UTF-8), 10,5 arrived as 105 (decimal separator), select values arrived blank.

Check: open three products from the run.

Fix: save as CSV UTF-8, check the decimal separator, re-run. Matched SKUs update, so the second run repairs the first.

4. Everything imported, and all of it is live

Or the reverse: everything imported and none of it visible. Both come from the setting Publish new products immediately. On (the default), every new product is live in every channel the moment it lands; off, new products arrive switched off.

Check: the Enabled figure at the top of PIM › Products.

Fix: if the run should not be live, select its products and disable them, then turn the setting off. If it should be live, enable the set deliberately. A row that states enabled explicitly wins either way.

5. Imported, but completeness says nothing

The products sit in the Never measured or Without a family queue: the file had no family column, or the code was wrong. A product without a family has no required attributes, so its completeness stays empty. The setting Default product family applies a family to products created without one, but an unknown code there is ignored without an error, so check its spelling.

Fix: select the queue, Assign family; the family is set and completeness measured in one step. If the family is right and the figure still empty, Recompute completeness.

6. Imported, but not in its categories

Category assignments are a separate step in the import order, or come from a rule. Rules run nightly and on Recompute rules, so a product imported at 10:00 is filed at three the next morning unless you recompute.

Fix: Recompute all rules on PIM › Categories, or Add to category on the selection.

7. A recurring feed overwrote enriched data

The nightly ERP feed replaced 200-word descriptions with 40-character invoice text, and the import reports success. The fix is a decision: settle which system owns which field and map only those fields. Recurring feeds run in Integration Studio; Integrations › Runs lists every run, see Monitor your syncs.

Still nothing

Capture these before you contact support:

  • The first ten rows of the file, including the header
  • The three result numbers
  • One failed row with its reason
  • One SKU that should have matched, and how it appears in the grid

Then contact support.