Which sheet. By default LedgerPair opens the first visible worksheet, so a hidden sheet at the front of the workbook stays out of the way. When a workbook has more than one sheet, a picker under the file lets you choose another, and the comparison runs on that sheet. The Summary sheet of the export records which worksheet each side used.
Where the header is. The first non-empty row is the header. Blank rows below it are skipped while their row numbers are kept, so a row number in LedgerPair is the row number you see in Excel. Empty columns at the end of the header are dropped.
Cell values, not cell formats. Numbers are read as stored values, dates as calendar dates, TRUE and FALSE as text, and formulas by their last calculated result. A figure formatted with two decimals in one workbook and none in the other compares as the same number, and a date typed in one notation matches the same date typed in another.
Files that are open. A workbook can stay open in Excel while LedgerPair reads it. The file is opened for reading with sharing allowed, and nothing is ever written to it.
Pairing and comparing. Rows pair on the key columns you tick, and every other column with the same heading in both files is compared after spaces, numbers, and dates are normalized. Headings match regardless of case and extra spaces, so Unit Price and unit price line up.
Reading the result. Each changed row says how many cells differ and in which columns, such as 2 cells differ: Price, Qty. Selecting it fills the detail pane with every compared column from both workbooks, with the changed cells highlighted. Rows found in only one workbook, and keys that repeat inside a workbook, appear in the same table, and the filter shows one kind at a time.
The export. Export evidence workbook writes a new .xlsx file. Summary carries the counts, the keys, and the columns compared. Additions and Deletions carry every column of the unmatched rows. Changes has one row per changed cell, with the value from each file and the normalized form that was compared. Duplicates lists each repeated key with its rows. Those four sheets have a frozen header row and filters, so the record is easy to sort and hand on.