Review two invoices with repeated item codes

Where the extra work starts

Retain the invoice identity alongside the line so your workbook can distinguish a repeat purchase from a duplicate.

Upload a representative supplier invoice, inspect its extracted fields and line items beside the PDF, then download Excel. Put the awkward invoice in your trial too: the useful result is less total work after corrections.

A duplicate extraction is the same source occurrence represented more than once. A repeated purchase or separate delivery can share a code while remaining a distinct source occurrence. The reviewer needs enough evidence to tell those cases apart.

A practical way to get it right

Assign each source file a stable reference and retain the printed invoice identifier separately. For each extracted row, record its page and line reference or another reliable source location. Use that composite identity when checking whether the same occurrence was processed twice.

Compare suspected duplicates across all relevant source fields and locations. If two rows point to the same source occurrence, investigate repeated page ranges, appended exports or reruns. If they point to different occurrences, retain both unless the authorized business review establishes another reason to change them.

Keep the raw extraction before any deduplication. Record excluded duplicate representations with the surviving reference and reason. Do not use spreadsheet “remove duplicates” on the item-code column alone, since that discards legitimate repetitions without explaining the decision.

OpenPeppol: Invoice line syntax defines a line identifier within its invoice. Preserve document context with that identifier instead of treating it as globally unique.

What this looks like on a real task

Synthetic example. The people, references and figures below are invented to demonstrate the method.

Three output rows represent two distinct source occurrences. The code is identical in all rows, but only the third is a duplicate representation in this synthetic example. Copy the source-occurrence, code and treatment columns. Preserve quantities and amounts too, while remembering that even identical values on different source lines can be legitimate.

Catch the cases that create rework

A revised invoice may reuse a number and item codes while replacing earlier evidence. That is a version question, not a simple duplicate-code test. Link the revision and ask the reviewer which document is current before combining its rows.

If source references were lost during export, recover them before destructive deduplication. A plausible row count after deletion does not prove that the remaining records are the right ones.

OpenPeppol: Item information describes item identifiers separately from line identity. A shared item code therefore does not by itself establish that two source occurrences are duplicates.

Decide whether the change is worth keeping

Reconcile accepted rows to unique source occurrences and retain an exclusion log for duplicate representations. Check that legitimate repeated codes survive sorting, merging and import. The resulting register should let another reviewer reproduce the deduplication decision without relying on the author's memory of which file was processed first.