1. Load the two files
Delimiter detection is a heuristic and can be wrong on a genuinely ambiguous file (European decimals, for instance). If the parsed column count looks wrong, override it above.
Pull the same report twice — a month apart, or before and after a migration — and find out what actually moved.
A generic text diff compares line 41 to line 41. Export the same report twice with a different sort order and every row shifts, so you get thousands of differences that are not differences.
DeltaCSV matches rows by a key column you choose — order id, SKU, email — and matches columns by name. Re-sort the rows, reorder the columns, and the result is still zero changes. Only the rows that genuinely changed show up.
Handled honestly, because these are the places a diff tool quietly lies: duplicate keys are never resolved by taking the first match, rows with the wrong field count are never silently dropped, and columns that exist in only one file are reported as column changes rather than counted as thousands of cell changes.
Delimiter detection is a heuristic and can be wrong on a genuinely ambiguous file (European decimals, for instance). If the parsed column count looks wrong, override it above.