How should you prepare CSV data before importing it?
Short answer
Keep an original copy, confirm the delimiter and header, inspect empty and numeric fields, and identify columns that must remain text. Apply only the cleanup your receiving system requires, then compare row counts and sample records. Test the actual import settings for leading-zero identifiers, long account numbers, dates, and accented text before relying on the result.
Profile first and transform second. Empty cells, spaces, and zero are different values. Removing spaces may be appropriate for names but inappropriate for a fixed-width code.
Which CSV checks should you repeat after cleanup?
Compare column names, data-row counts, selected identifiers, and known amounts. Deduplication should reduce the row count by the number of removed records. Sorting should preserve it. Splitting should preserve the sum of data rows across parts.