Before importing a repaired product CSV, ask for the change log

A repaired CSV should come with an explanation of what changed. Otherwise, you are being asked to upload a file without knowing whether it fixed a formatting problem or silently guessed a product fact.

Keep the original export. Before importing any proposed replacement, review a change log that lets you trace a changed field back to the same product and its evidence.

This is an illustrative workflow, not a customer case study or a promise that every field can be repaired. Prepared with AI assistance by Rowmint, the product-data education brand of Mingche Commerce.

Three rows, three different decisions

Example issue	Evidence available	Appropriate next step
A field contains accidental surrounding spaces	The receiving field's rules and the intended value are known	Propose a supported formatting correction; show old and new values
A product identifier is missing	No manufacturer record supplied	Request the source identifier or relevant product evidence; do not invent digits
The linked page opens a different size	The intended variant is confirmed, but no working variant link is known	Keep the link issue unresolved until the storefront mapping can be checked

The useful distinction is between a supported correction and a missing fact. The second category is a request for evidence, not an invitation to make the output look complete.

What to keep beside the proposed file

• Stable row ID: enough to find the same product in the original and proposed exports.
• Field name and before/after values: a changed-row count alone cannot explain an edit.
• Reason and evidence: what supports the proposed correction, including any merchant-supplied source.
• Unresolved items: what still needs information, who can supply it, and why it was left unchanged.
• Input version and check time: so an old proposal is not confused with a newer catalogue export.

Review one row before importing many

Compare the proposed values with the current source data. If inventory or prices have changed since the export, refresh those facts before applying the file. Then use the import tool's preview or validation flow where available. Check a small, appropriate sample after import and retain the original file and change log.

Importing a CSV can overwrite existing values. A downloadable repair proposal and an update to your live shop are separate steps. Neither establishes approval by Google or another platform.

Check the file structure before judging the repair

If a file opens as one wide column, first confirm how the receiving tool reads its separator. If two rows appear to describe the same item, confirm their stable product and variant IDs before merging or deleting either row.

• Check CSV delimiters and column boundaries before treating a parsing problem as a product-data correction.
• Investigate duplicate product IDs while preserving the original rows and evidence.

These free guides help you review a file. A file that parses correctly is not proof that its prices, stock or identifiers are correct, and reading a guide does not change your live store.

At Mingche Commerce, the scope of supported CSV repairs matters as much as the output file. Review the listed coverage and any request for missing information before purchasing.

Start with a free product-page check, then compare audit and repair scope.