Designing a CSV Import Repair Preview

A CSV import repair preview should make the product helpful before it becomes strict. The goal is not to accept bad data. The goal is to show the user exactly what will happen before anything durable is written.

Start with a temporary upload stage. Parse the file, detect headers, count rows, and show a small sample. Do not create records yet. This stage should tolerate blank lines, renamed columns, extra columns, and common delimiter issues because the user is still in preview mode.

Next, provide mapping with confidence. If “Email”, “email_address”, and “Contact Email” all look like the same field, suggest the mapping but show it as a suggestion. If two columns could map to the same required field, force a choice. If a required identity field is missing, explain why the import cannot commit until the user chooses a stable identifier.

Then group errors by repair action. Users should not read one thousand row errors. Group them as missing required value, unknown status, invalid date, duplicate identity, destructive update risk, or ignored extra column. Show the affected row count and a few examples. Let the user fix a column rule once rather than edit every row manually.

Finally, make commit strict. The final step should show records to create, records to update, rows to skip, and fields to transform. Users should be able to download the rejected rows. The product should not silently guess identity, overwrite important records, or coerce values that change meaning. Helpful preview plus strict commit is the core design.

Designing a CSV Import Repair Preview

// COMMENTS

ON THIS PAGE