Pull a PDF's tables into a spreadsheet

How to convert PDF to Excel

1. Add your PDF. Drag the file onto the drop zone. Nothing is sent anywhere — the reading happens on your device.
2. Narrow it to the pages you need. A 90-page report with one table on page 41 converts better if you say so. Type 41 in the page range.
3. Choose what to include. “Detected tables only” keeps the rows with real column structure and drops the prose around them. “All text” takes every line.
4. Convert and check the preview. The first rows are shown as they will open. Figures shown right-aligned came through as numbers.

A PDF does not contain a table

This is the thing worth knowing before judging any converter, this one included. A PDF has no table object. It has no rows, no cells and no column definitions. What it has is a list of text runs, each with a position on the page, and the fact that some of them line up into what your eye reads as a grid is a coincidence of typesetting rather than a structure recorded in the file.

Quite often the ruled lines you can see around a table are not connected to its text at all — they are vector strokes drawn separately, in a different part of the content stream, by a program that had a table and threw the structure away when it exported.

So converting a PDF to a spreadsheet means inferring the grid back from geometry: grouping text into rows by where the baselines sit, then finding the vertical channels of whitespace that run down the page and treating those as column boundaries. That inference is the product. It is also why every honest converter, at any price, gets some tables wrong.

Why the numbers arrive as numbers

A column of figures that lands in Excel as text is only half a conversion. It looks right, sorts wrongly, and refuses to sum — and the repair, selecting the column and running Text to Columns, is exactly the manual work you came here to skip.

So values that are unambiguously numeric are written as numbers, with the formatting they arrived in preserved: a thousands separator is stripped and remembered, a currency symbol becomes a cell format, a percentage becomes a real percentage, and an accounting negative written in parentheses becomes a negative number.

The parsing deliberately stops short of clever. Anything ambiguous stays as text, because a figure quietly changed is far worse than a figure left alone — a part number like 10,000,00 is not ten thousand, and a converter that decides otherwise has damaged your data while appearing to succeed.

When this will not work, and what to do instead

If the PDF is a scan, there is no text to read and the conversion returns nothing. A photograph of a table is pixels; the tool will tell you so rather than hand back an empty sheet. Run OCR first and the text layer it adds becomes readable here.

If the table has no gap between two columns — a common result of a narrow layout with right-aligned figures butted against the label beside them — those two columns merge into one. Widening the page range does not help, but converting with “All text on the page” and separating the column in the spreadsheet often does.

If a table runs across several pages with a repeated header on each, choose “Everything on one sheet” so the parts land in one continuous range instead of being scattered across a sheet per page.

Doing this without uploading a financial document

The PDFs people convert to spreadsheets are bank statements, invoices, payroll summaries and management accounts. They are, almost by definition, the documents least suited to being posted to a stranger's server for processing.

Everything here runs in the page. The file is read with the File API, the text extraction and column detection run in your browser, and the workbook is assembled in memory before being handed to your downloads folder. Nothing is transmitted, so there is no copy to retain, no retention policy to read, and nothing to delete afterwards.

Frequently asked questions