What PDF to Excel does
A PDF does not store a table as rows and columns. It stores pieces of text, each with a position on the page, plus lines and boxes that are drawn around them. This tool reads those pieces of text, works out which ones sit on the same line and which ones line up below each other, and writes the result as an Excel workbook (.xlsx) or a CSV file. Values that look like numbers become real numbers, so you can sum, sort and chart them right away.
It works on PDFs that were made by a program: bank and card statements, invoices, price lists, reports and exported tables. For a scanned PDF, run OCR PDF first, which adds a text layer, and drop the result here. If you only need the text, PDF to Text is quicker, and to edit the whole document, try PDF to Word.
How rows and columns are found
The PDF is opened with pdf.js, Mozilla's open source PDF library, which loads only when you drop a file. For every page it returns the text pieces with a transformation matrix, from which the tool takes the position of each piece, its width and its font size.
- Rows: pieces whose baselines are less than half a font size apart form one row. The larger text of a line sets its baseline, so a small superscript stays in its row.
- Cells: inside a row, two neighbouring pieces end up in the same cell when the space between them is smaller than the column gap times the font size. A space is added between them when the gap is wider than 0.15 font sizes, because many PDFs do not store the space between words as a character.
- Columns: the tool looks at every row with at least two cells and marks the stretch of the page each cell covers. Where these stretches overlap, they belong to one column; every strip of the page that none of them covers separates two columns. Lines with one cell, such as a title, are left out of this step and put in the column where they start.
With All pages in one sheet, the columns are found across all chosen pages together, so a table that runs over many pages keeps the same columns from start to end. When most of the text on a page runs sideways, as in a landscape table on a portrait page, the page is read turned so that this text runs left to right. Text in other directions is left out, and the tool says how many pieces that were.
What turns into a number
| In the PDF | In Excel |
|---|---|
| 1,234.50 | Number 1234.5 |
| -12 or −12 | Number −12 |
| (1,250.00) | Number −1250, the accounting style for negatives |
| 1.234,56 or 1 234,56 | Number 1234.56 with the 1.234,56 setting |
| 12.5% | Text, or 0.125 with Percentages as numbers |
| $19.99, 01/03/2026, 00501 | Text |
| 4111111111111111 | Text: more than 15 digits |
Numbers with a leading zero and numbers with more than 15 significant digits stay text on purpose. Microsoft notes that Excel removes leading zeros and keeps at most 15 significant digits, so a ZIP code or a card number would lose digits as a number. A percentage turned into a number is stored as a fraction, so 12.5% becomes 0.125. The CSV file holds the cell text as it appears in the PDF, in UTF-8 with a byte order mark, and the rows of all chosen pages one after another, since a CSV file has no sheets.
Worked example
Press Try a sample PDF. It is a statement set in 10-point Helvetica: dates at 72 points from the left edge, descriptions at 160, and amounts right-aligned so they end at 420. The space between Coffee and beans is 2.78 points wide, less than one font size (10 points) at the default column gap of 1.0, so Coffee beans is one cell. The date 01/03/2026 is 50.04 points wide and ends at 122.04; the gap to the description is 37.96 points, almost four font sizes, so a new cell starts there. Amounts such as 12.50 and (1,250.00) start at different points but overlap at their right ends, so they share one column. The result has three columns, Date, Description and Amount, and the amount (1,250.00) arrives in Excel as the number −1250.
Limits
- Scans and photos have no text to read. The tool tells you which pages have no text; run OCR PDF first.
- Merged cells are not detected. A heading that spans several columns lands in the column where it starts.
- Cells with several lines become several rows, because each line has its own baseline.
- Rotated text that runs against the main direction of the page, and tables drawn as images, are not read.
- Lines, colors, fonts and formulas are not carried over; the workbook holds plain values.
- Encrypted PDFs that need a password to open cannot be read. Unlock your copy first with Unlock PDF.
How to convert a PDF table to Excel
- 1Drop your PDF on the box. The tool reads the text of every page and shows how many pages have text.
- 2Choose the pages, one sheet per page or all pages in one sheet, and how numbers are written in the PDF.
- 3Check the preview. If columns run together, lower the column gap; if a cell splits, raise it.
- 4Press Download Excel (.xlsx) or Download CSV.
To turn a CSV file you already have into a workbook, use CSV to Excel.
Sources
- pdf.js source, api.js: getTextContent returns text items with str, dir, transform (transformation matrix), width, height, fontName and hasEOL
- pdf.js API, PDFPageProxy.getTextContent: whitespace in the text content is replaced by standard spaces
- Microsoft Support: Keep leading zeros and large numbers: Excel removes leading zeros and has a maximum precision of 15 significant digits
Related guides
- CSV Shows Wrong Characters in Excel: The UTF-8 BOM Fix
- Excel's 1,048,576 Row Limit and How to Open a Bigger CSV
- When Does a Work Enter the Public Domain in the US?
Frequently asked questions
Is my PDF uploaded?
No. Your browser reads the text of the PDF, builds the table and writes the Excel and CSV files on your device. Nothing is sent to a server.
Why does the tool say my PDF has no text?
The PDF is probably a scan or a photo: each page is a picture of the paper, not letters. Run it through OCR PDF first, which adds a text layer, then drop the new file here. The OCR text is only as good as the scan, so check the numbers afterwards.
Which values become numbers in Excel?
Values like 1234, 1,234.50, -12, 12.5 and (1,250.00), which is read as minus 1,250. With the 1.234,56 setting, a period or a space separates thousands and the comma marks decimals. Percentages become fractions only when you tick that box (12.5% becomes 0.125). Currency signs, dates, values with a leading zero like 00501 and numbers with more than 15 digits stay text.
Two columns came out as one. What can I do?
Lower the column gap. The tool starts a new cell only where the space between two pieces of text is at least as wide as the gap setting times the font size. Tables with narrow spacing need a value like 0.5. If words of one cell are split into two columns, raise it instead.
Does it keep merged cells, colors and borders?
No. The tool reads only the text and where it sits on the page. Lines, colors and fonts are not read, merged cells are not detected, and a table that is drawn as an image has no text to read at all.
What is the difference between one sheet per page and all pages in one sheet?
One sheet per page gives every page its own sheet, named Page 1, Page 2 and so on, with columns found on that page alone. All pages in one sheet puts the rows of every chosen page one below the other and finds the columns across all of them, which suits bank statements and reports where a long table runs over many pages.