Validate a workbook's row count after manual merging

A total is necessary but not sufficient

Reconcile counts and source coverage so the combined workbook is complete enough for the next task.

ParseMill exports reviewed invoices to Excel with separate summaries, line items, sources and review status. You can import an XLSX or CSV template and save a client setup. Run a real invoice through to the final workbook and check it in the application that receives your work.

Define the unit being counted. A source line, a table row and a reconstructed observation may differ when descriptions wrap or records continue across pages. Use the same unit throughout the reconciliation.

Use a merge reconciliation sheet

The duplicate count is valid only if the two removed rows represent duplicate copies of the same source observations. Equal sample codes or equal results are insufficient. Retain a removal log with the kept and removed keys so the final count can be reconstructed.

Microsoft: Table.RowCount provides row counts for table reconciliation. Pair those counts with identity checks to establish which observations survived.

Reconcile after the merge

1. Inventory each input version, reviewed scope and expected record keys. Exclude headings and summaries under a documented rule before counting observations.
2. Merge complete records including source identity and review states. Retain an input-origin field in the destination.
3. Compare the destination key set with expected contributions. List missing, repeated and unexpected keys separately before deciding any repair.
4. Apply only source-supported removals, then recompute counts and check records around each manual insertion or deletion. Save the reconciliation with the merged revision.

Microsoft: Table.Distinct removes duplicates according to selected comparison criteria. Choose those criteria deliberately; a result-only or sample-only comparison can erase legitimate separate observations.

Investigate mismatches without forcing the total

If the destination has twenty-five rows instead of twenty-six, do not add a blank row to make the count agree. Identify the missing observation from the key comparison. If it has twenty-seven, locate the extra key or duplicated copy rather than deleting the last visible row.

A matching total also needs investigation when the key set differs. One omission and one duplicate cancel numerically while leaving the register wrong. Keep the identity reconciliation as part of acceptance. Check records from every contributing file, including the first and last record copied from each source range.

Verify the resulting population

For the fixture, the final register contains twenty-six observations with an explained source contribution and two documented duplicate-copy removals. The recipient can trace each retained record and understand every exclusion. Counts and keys should both agree with the merge log.

When combining ParseMill exports manually, use this reconciliation before sharing the workbook or loading it into another system. It turns a visual copy-and-paste operation into a reviewable record transformation.