PDF to Excel
Drop a PDF and download an .xlsx workbook. Each page becomes a sheet with detected tables and surrounding text preserved.
How it works
- Drop a PDF into the upload area. The engine works on any modern PDF with selectable text.
- Each page becomes one sheet in the output workbook. Detected tables turn into proper Excel ranges with bold header rows.
- Cell values are auto-typed where the heuristic is confident: numbers (with thousands separators / accounting parens), dates (ISO, US, named months), and booleans.
- Click 'Convert to Excel'. The full pipeline — text extraction, line clustering, column detection, table inference, type coercion — runs locally.
- Download the .xlsx and open it in Excel, Google Sheets, LibreOffice, or Numbers.
Frequently asked questions
Are table cells actually typed (numbers, dates)?
Yes — the heuristic detects '1,234.50' / '-42' / '(123.45)' / '15%' as numbers, ISO and US-format dates, and yes/no/true/false as booleans. Anything ambiguous stays as text. Column-uniform typing locks a column to its first detected type so a stray text row in a numeric column doesn't break things.
What about tables without visible borders?
We fall back to column inference from text x-alignment when no rule lines are detected. Works for most aligned text-grids — though very dense or jittered layouts may merge or split columns.
Is non-table text preserved?
Yes. Headings render as bold rows above the relevant table, body paragraphs as text rows, all in PDF reading order so invoices keep their vendor block, date line, etc. next to the line-item table.
Does it handle scanned PDFs?
No — scanned PDFs are images with no text layer. Run an OCR tool first to make the PDF searchable, then convert.
Are formulas reconstructed?
No. PDFs don't carry formula info; we extract the displayed values verbatim. You can recompute in Excel after import.
Is my PDF uploaded?
No. The converter runs entirely in your browser tab — no upload, no server.