Convert from PDF · guide
What PDF to Excel does
PDF to Excel pulls the tables out of a PDF into an .xlsx workbook, one sheet per table, joining a table that runs across a page break back into one. Figures such as $18,430.50, (2,100) and 12% become real numbers you can sort and add up. Tables the tool is not sure it has read correctly are left out and reported rather than guessed. The PDF is never uploaded.
How to convert a PDF table to Excel
- Choose a PDF with tables Select a PDF that contains tables, such as a statement, invoice or report exported from another program. A scanned PDF has no text to read, so run it through OCR PDF first, and a password-protected one through Unlock PDF.
- Extract the tables Select Extract the tables. Each table the tool can identify with confidence becomes its own sheet.
- Check and save Read the notes, which say how many tables were found and which were left out, then save the workbook and spot-check a few figures against the PDF.
Worked example
Extract a ruled invoice table for reconciliation.
- Starting point
- Choose a PDF whose item rows, column headings and totals are clearly separated.
- What to do
- Select Extract the tables, then compare each sheet’s rows and columns against the page before calculating anything.
- Result
- Expect one sheet for each table it finds, named after its page (Page 1, for example), with amounts stored as numbers. A grid it is not sure of is left out and the notes name its page. If the invoice table carries on from page 3 to page 4 with the same columns in the same places, expect one sheet named Pages 3–4, with a repeated header row dropped, and (2,100) read as −2,100. Verify the item count, the decimal places and the final total against the PDF.
If the result is not right
If columns merge or totals shift, correct the spreadsheet against the original or use a cleaner source. Never rely on a total without checking the component rows.
Limits worth knowing
- Merged cells, borderless tables and irregular spacing are ambiguous; conversion is not an accounting validation.
Questions and answers
Why is a table from my PDF missing from the spreadsheet?
Because the tool was not confident it had read it correctly. A PDF has no real tables, only text placed so that it lines up, and a wrongly split column produces numbers that look right and are not. Rather than risk that, every table the tool finds gets a confidence score from 0 to 1, any table scoring under 0.7 is left out, and the result says on which pages. Tables drawn with ruling lines come through best, since the lines are direct evidence of the grid.
Can I use the numbers in formulas?
Yes. Amounts such as $18,430.50, bracketed negatives such as (2,100) and percentages such as 12% are written as real numbers, so you can sort, filter and sum them; 12% is stored as 0.12 and shown as 12%, but a percentage that is not a whole number, such as 12.5%, is stored as 0.125 and shown as 0.125, so give that column a percentage format with one decimal place to see 12.5%. Dates written as 2026-03-14 become real dates. The currency symbol is dropped from the cell, so set a currency format on the column if you want it shown. Anything that does not read cleanly as a number is kept as text, where it is easy to spot.
Why are some dates and account numbers left as text?
On purpose, where converting is most likely to change the value. A date like 03/04/2026 is the 3rd of April in most of the world and the 4th of March in the United States, and nothing on the page says which. Numbers with leading zeros, more than 15 significant digits (Excel keeps only 15, according to Microsoft's Excel specifications and limits), or 12 or more digits with no separators stay text so no digit is lost, as does anything in a column headed as an account, ID, reference or phone number. A lone comma is read as a thousands separator only when one to three digits come before it and exactly three after, so 1,085 becomes 1085, even where the document means 1.085; any other lone comma, as in 12,5 or 12,50, leaves the number as text. A single dot is always a decimal point, so 1.234 stays 1.234. A number with both, such as 1.234,56 or 1,234.56, takes whichever comes last as the decimal point. If your document writes numbers the continental way, check those columns against the PDF.
Why does my bank statement say it could not be opened?
Most often because the statement is password-protected, as many banks send them. A locked file cannot be read: open it in Unlock PDF with the statement's password, then convert the unlocked copy. Both steps run in this tab, so the statement is not uploaded. If the message says the PDF has no text layer instead, the statement is a scan or a photo, and OCR PDF has to add the text before any table can be found.
How are the tables arranged in the workbook?
One sheet per table, named after where it was found: Page 4, Pages 3–5 for a table that ran across a page break, or Page 4 table 2 where a page holds more than one. A continued table is joined into one sheet and its repeated header row dropped. A header row is set in bold, shaded and frozen so it stays in view, columns are sized to their contents, and a sheet with six or more columns is set to print in landscape.
Where did the rest of the text go?
Only tables are written to the workbook. Paragraphs, headings, footnotes and images are left out, and the notes say so, because a sheet is for rows and columns. Text printed sideways, such as a rotated column heading, is dropped too, which can leave a header cell empty. For the whole document as editable text, use PDF to Word; for plain text, PDF to Markdown.