Matched / unmatched separation
Cleared rows are not the default work list.
Exception management
Instead of checking every row, users work through an exception queue. Matched rows stay available. The workbench is for unmatched, duplicate, tolerance, and other review states.
A 50,000-row match is not a 50,000-row review. The operational benefit is that people spend time on the exceptions, with a status, an owner, and a comment thread, instead of coloring cells in the source file.
Sample preview
Conceptual comparison. Not a live count.
How it works
Step 1
Matched and unmatched are stored from the run.
Step 2
Filter to the exception types that still need a person.
Step 3
Give an item an owner and a note.
Step 4
Move items through review states. Use reason codes where supported.
Step 5
Comments and status changes remain on the exception.
What this workflow is designed to do with uploaded files.
Cleared rows are not the default work list.
Each item has a review state so the queue has a definition of done.
Notes stay on the exception instead of in a side email.
An owner can be recorded so open items are not anonymous.
Where supported, a reason code makes later reporting consistent.
Items can sit in review until a person with permission closes them.
The trail of status and comments is part of the record.
After a vendor run, 15 exceptions remain. An analyst assigns the missing invoice to themselves, comments that the PDF is requested, and leaves the item open. A reviewer later sees the same thread — not a new email chain.
Exception management is how reconciliation becomes operations instead of a one-off spreadsheet. The queue is the work. The matched file is the evidence.
Continue with the workflows that usually sit next to this one.
No. They remain available. The default workbench focus is exceptions.
Yes. Comments and assignment are part of the exception record.
No. Humans remain in control of status and approval.
Run a reconciliation, then review only the rows that did not match.