Reconciling a statement
The example compares a bank statement with a ledger export by transaction reference. T-03 differs by 10.00 and shows as changed. T-04 exists only in the bank file and T-05 only in the ledger. T-06 is posted twice in the ledger: rather than pick one, the comparison reports the duplicate key and leaves it out of the field-by-field diff.
T-01 differs by 0.004. With the numeric tolerance set to 0.01 it counts as unchanged, and the summary says how many values matched only within tolerance, so the leniency is never hidden.
Positional mode compares row 1 with row 1 and so on; use it only when both files share a stable order.
Questions
Which columns are compared?
Columns with the same header name on both sides. Columns that exist on only one side are listed separately.
Can I compare a file with my cleaned result?
Yes. Choose “Recipe result” as A or B to compare the output of the recipe with an original or with another file.
How do I export the differences?
Use “Download comparison report”, a CSV with one line per difference, duplicate key and one-sided row.