Reconstruct table order after pages were assembled incorrectly

Separate received order from intended reading order

Reconstruct the sequence before combining rows so the result preserves the relationships the report intended.

For multipage invoices, use ParseMill to review extracted invoice data beside the source and export separate summary and line-item sheets. Include a page break in your trial and check that every required line made it through. Other report types need their own fit assessment.

Printed labels can help, but covers, appendices and restarted numbering complicate a simple numeric sort. Use report identity, section title, continuation markers and neighboring record ranges together. A proposed order should remain explicitly proposed until the sequence is supported or the source owner confirms it.

Build the order worksheet before rearranging anything

Adobe: Renumber pages in PDFs documents page labels. Microsoft: Table.AddIndexColumn can preserve imported row positions, which helps relate a reordered working table back to the original output.

1. Create fields for physical position, printed label, report/section, first record, last record, proposed position and decision evidence. Leave proposed position blank when uncertain.
2. Identify covers and section dividers separately from detail pages. Keep their role even if they contribute no extracted rows.
3. Check continuity across each proposed boundary. Look for missing or repeated identifiers, a carry-forward value, a sentence continuation or a stated end of section.
4. After review, create a working order map. Reorder a copy only if needed and retain the map with the unchanged received file. Label source references as received-position or working-position.

Use two coordinates for the same page

Synthetic example: an invented report was assembled with the second detail page before the first.

The working sequence is received positions 1, 3, 2, 4. The appendix does not move beside Page 1 merely because its printed label contains the same number. An extracted row for R16 should retain received position 2 even if it is displayed after R10 in a reordered register. Store the working position as a separate reference so a reader can locate the evidence in either copy.

Stop where the sequence is unsupported

A jump from R10 to R21 may mean a missing page, a deliberate gap in identifiers or an unrelated section. Do not invent R11–R20 to make the sequence look complete. Similarly, sorting timestamps can be misleading when a report groups records by site, category or sample rather than chronological order.

If two candidate pages share a printed label, compare report identifiers and revision details before choosing an order. Preserve conflicting evidence in the worksheet. A source owner can resolve the intended assembly; an extraction tool cannot reliably infer it from a familiar heading alone.

Verify the copy and communicate the mapping

Check that the reorder map contains each retained received position exactly once. Compare the first and last row at every changed boundary, then verify several source links against the original PDF. State which coordinate system the final workbook uses and provide the crosswalk to authorized recipients.

For the fixture, all four received positions remain represented, and only the order of positions 2 and 3 changes. A complete map does not prove that the source has no omitted pages. Use the multipage review checklist to assess both sequence and completeness before relying on the register.