loading tool…
What PDF to Excel does
PDF to Excel extracts tabular data from a text-based PDF into an editable .xlsx workbook, inferring rows and columns from the position of the text.
This is the hardest extraction on the site and the honesty is the whole point. A PDF has no concept of a cell — it stores glyphs at fixed x/y positions. Turning that back into a spreadsheet means inferring both the rows and the columns from where the text sits, which only really works when the PDF is genuinely tabular: data laid out in aligned columns with clear gaps between them.
Your PDF is read in the browser with pdf.js. Text is grouped into rows by its baseline, and columns are found by looking for the vertical bands of whitespace that separate them — a method that copes with both left-aligned labels and right-aligned numbers. Each page becomes its own sheet in the workbook, written as real editable cells with ExcelJS.
How to use PDF to Excel
Upload a PDF
Drop in a text-based PDF with tables. Scanned PDFs have no text to extract — OCR them first.
Convert
Click Convert to Excel. Each page is scanned for rows and columns, with progress shown.
Download and check
The .xlsx downloads automatically. Open it and check the columns landed where you expect before relying on the data.
PDF to Excel: details and limits
How well it works depends entirely on the PDF. Cleanly aligned tables with gaps between columns come out well; borderless, merged or nested tables come out imperfect, multi-table pages can run together, and a page of prose becomes a single column of lines.
You get the text values in a grid — numbers arrive as text you can convert in Excel — but formulas, number formatting, colours, fonts and merged cells are not recovered. The point is to get the data out to work with, not to reproduce the styling.
A scanned PDF has no text layer, so there is nothing to extract; the tool stops and points you to OCR PDF, and reports any pages that had no selectable text. Very large pages are trimmed to 5,000 rows and 50 columns, and truncation is reported rather than silent.