Retyping figures out of a PDF is slow and error-prone, and a single transposed digit in a financial table can go unnoticed for months. Upload your PDF above and get an editable .xlsx file instead — numbers in cells, ready to sort, filter and calculate.
This is the hardest conversion on the site, and it is worth understanding why before you start. What follows is what actually determines whether your document converts cleanly, and how to check the result properly.
A spreadsheet knows it is a grid. Every value belongs to a cell with a row and a column, and that relationship is stored explicitly.
A PDF stores none of that. It records that a piece of text sits at particular coordinates on the page, in a particular font, at a particular size. The neat table you see is an illusion created by careful positioning — there are no rows, no columns, no cells, and often no lines either.
So conversion means reconstructing the grid by inference: reading where every value sits, spotting the vertical alignments that suggest columns, the horizontal ones that suggest rows, and deciding which text is a header and which is data.
This works remarkably well on ordinary tables. It struggles exactly where a human eye would also hesitate — ambiguous spacing, values that straddle two columns, rows of different heights.
Knowing where your document sits saves disappointment.
Converts cleanly: tables with visible ruling lines or clear consistent spacing, one row per record, single-line cells, a plain header row, and consistent column alignment. Bank statements, invoice line items, price lists and exported reports usually fall here.
Needs checking: tables without ruling lines, merged or spanning cells, multi-line entries inside a single cell, tables that continue across several pages with repeated headers, and footnote markers attached to values.
Will not reconstruct properly: financial statements laid out as designed pages rather than tables, documents where columns are separated only by variable whitespace, nested sub-tables, and anything where the visual grouping comes from shading rather than position.
A quick test before you convert: open the PDF and try to select one column of figures by dragging down it. If the selection follows the column, the underlying positions are clean and conversion should be good. If it grabs whole rows or jumps around, the text order in the file does not match the visual layout, and the result will need work.
Try selecting a number in your PDF. If nothing highlights, the page is a photograph and there is no text inside to extract — conversion will produce an empty sheet.
Scanned tables need OCR PDF first, which recognises characters from the image. Two cautions that matter more with numbers than with prose:
OCR confuses visually similar characters, and in a table of figures that is dangerous rather than merely untidy. The classic pairs are 0 and O, 1 and l and I, 5 and S, 8 and B. A misread letter inside a number turns the cell into text, and every formula referencing it silently fails or returns zero. Always spot-check totals against the original.
Straighten and orient the scan first. Recognition accuracy drops sharply on rotated or skewed pages — run Rotate PDF before OCR, not after.
This is the step people skip, and it is the one that causes real damage.
Values can arrive in Excel as text that looks like a number. When that happens, SUM ignores them, sorting puts 100 before 20, and totals come out wrong without any error message. Excel usually shows a small green triangle in the corner of such cells, but not always.
Three common causes, all worth checking:
$1,234.56 may import as text rather than the value 1234.56.1.234,56 means one thousand two hundred in much of Europe and something quite different elsewhere. Check which convention the source used.(1,500) for −1500. Excel reads that as text unless converted.The fastest verification: select a column of figures and look at the status bar at the bottom of Excel. If it shows a Sum, the cells are numeric. If it only shows Count, they are text. Data → Text to Columns, then Finish, converts a whole column in one step.
A table running across twelve pages usually repeats its header row on each one, and may carry page numbers or footers between sections.
After conversion those repeated headers become ordinary data rows scattered through your sheet, which breaks sorting and filtering. They are easy to remove once you know to look: sort by the first column and the duplicated headers group together, or filter for the header text and delete those rows in one go.
The same applies to subtotal rows. If the original repeated running totals per page, those will now sit inside your data and will be double-counted by any SUM across the whole column.
Working with a scan? OCR PDF makes it readable first. PDF to Word suits documents that are mostly prose, PDF to TXT gives raw text with no structure to correct, and Excel to PDF converts back once you have finished. Browse all PDF tools.