Skip to content

PDF to Excel puts text in the wrong columns: why, and the fixes

By the getPDF team · Published 11 October 2026

The short answer

A PDF has no table inside, only text placed at positions, so every converter guesses columns from where the text sits and from the lines drawn on the page. Text lands in the wrong column when that guess is spoiled: a block of text crossing the gaps, columns that almost touch, or spaces instead of positions. Split stuck cells with 2 formulas, split a collapsed column with Text to Columns, and check with a sum.

Try the converter on your table

Try it here, nothing is uploaded

PDF · any size · many at once

How the converter finds columns

It puts every piece of text into a row by its height on the page. Then it finds the columns in 1 of 2 ways. If the table has vertical lines drawn between its columns, it uses them; the result card then says “Used the table lines for the columns”. If not, it looks for vertical strips of white space that run down through the rows, the way your eye finds columns, and puts a column break in the middle of each strip.

2 small tables side by side. In the left one, 3 columns of text have clear white strips between them, marked as column breaks. In the right one, a line reading Invoice no. 2026-0412 runs across the first strip, which is marked as lost, so Pos and Description end up in 1 cell.Gaps run through: 3 columnsPosDescriptionAmount01Logo design85002Business cards129breakbreak1 line crosses a gap: 2 columnsInvoice no. 2026-0412PosDescriptionAmount01Logo design85002Business cards129lostbreak
Why columns merge. Left: the white strips between columns run all the way down, so each becomes a column break. Right: 1 line of other text crosses the strip, so the converter sees no gap there and 2 columns share a cell.

The 4 layouts that go wrong, measured

We built invented test pages for each case and converted them. Tables that behaved well, for contrast: a plain grid with gaps, the same grid with lines between every column, a grid with only horizontal lines, a grid inside a drawn rectangle, a heading spanning 2 columns, and a column filled in 1 row out of 12. All of those came out right.

Layout What we got Fix
A text block above the table crosses the column gaps (an invoice’s number and date block) Pos and Description in 1 cell, “Unit price” heading 1 column left of its prices Fix 1
2 columns almost touching, 2 points apart “AT-4020-LINZLinz” in 1 cell, glued without a space Fix 1, by hand
Columns made of typed spaces in a normal font “Qty Price” and “2 4.80” in 1 cell, as text Fix 2
2 tables side by side whose rows are at different heights their rows interleaved, 1 from each table in turn Fix 3

A table with only an outer frame, 4 lines and none between the columns, converts right too, even under a letterhead wider than the frame: the converter sees that no line runs between the table’s text and finds the columns from the gaps. The same typed-space table in a fixed-width font such as Courier converted correctly, because there the spaces keep the columns apart. 2 side-by-side tables whose rows line up came out as 1 table of 4 columns, which is easy to cut in 2.

Fix 1: split a stuck cell with 2 formulas

When 2 columns share a cell, like “01 Logo design, 3 drafts”, and the first part has a fixed length, 2 formulas split them:

  1. In an empty column, =LEFT(A9,2) gives “01”.
  2. In the next, =MID(A9,4,200) gives “Logo design, 3 drafts” (start after the 2 characters and the space).
  3. Fill both down, copy them, and paste back with Paste Special, Values.

When a heading sits 1 column off, as “Unit price” did, cut the heading cell and paste it 1 cell to the right. Amounts are not affected: on our invoice they all arrived as numbers, in the right rows, and summed to the printed net total.

Fix 2: Text to Columns on a collapsed column

When a whole row is in 1 cell, Excel can split it:

  1. Select the column, then Data, Text to Columns.
  2. Choose Delimited, tick Space, then Finish (Microsoft’s help “Split text into different columns with the Convert Text to Columns Wizard”, checked on 11 October 2026).

This works when each cell is a single word or number, like “Notebook 120 1,440.00”. It also splits multi-word descriptions into several cells, so on a table with descriptions, expect to join a few cells by hand, or use Fix 4.

Fix 3: tidy interleaved tables

Sort is the quick way: 2 tables whose rows alternate have their cells in different columns, so select the block, then Data, Sort by the first column of table 2. Its rows gather at the top or bottom, and you cut them away to a sheet of their own. If the order of rows matters, add a column with 1, 2, 3 before sorting so you can sort back.

Fix 4: decide whether to retype

A table of 10 rows is often quicker to retype, and a retyped table leaves nothing to check. For 200 rows, repairing the conversion wins. The rule: if the first page needs more than 2 of the fixes above, retype that table, and keep converting the ones that come out clean.

Before you convert: spot trouble early

  • Lines between columns are the best case. A table that has them converts column for column; the result card says it used the table lines.
  • A table on a page of its own, or with clear space above and below, converts better than one sharing its width with an address block.
  • Pages box: convert only the table’s pages (type 12-14) to keep other pages out of the sheet. Columns are still worked out page by page, so this does not change the result on the table’s own page. To keep a PDF of just those pages, use Extract PDF pages.
  • Scanned tables have no text at all; run OCR PDF first, and expect more fixing, because a scan’s lines are part of the picture.

The honest part

No converter, ours included, reads a table out of a PDF: they all reconstruct one from positions, so every one has layouts it gets wrong. Ours does not undo columns that are glued together on the page. What it does do is keep every piece of text, so nothing is silently dropped, and write amounts as numbers you can check with a sum. Do that check on the first file of every layout, as described in Invoice PDFs into Excel; for long tables the seam checks are in Multi-page table to Excel.

Questions

Why did my PDF to Excel conversion put everything in 1 column?

Usually because the table's columns are separated by a few spaces typed in a proportional font, which hides the column breaks. Text to Columns in Excel splits the column again when each cell is a single word or number.

Why are 2 columns stuck together in 1 cell?

Either the columns almost touch on the page, or a line of other text above or below the table crosses the gap between them. On our test invoice, the block with invoice number and date did exactly that to the Pos and Description columns.

Does converting only the table's page help?

It keeps other pages out of the sheet, which is worth doing. It does not change the columns on the table's own page, because columns are worked out page by page.

Would another converter do better?

Sometimes, on a particular layout. Every PDF to Excel converter has to guess columns from positions, because a PDF stores no table. If a table matters, check the first result from any tool with a sum against the printed total.

The tools for this job