PDF to Excel

Pull the tables out of a PDF into a real spreadsheet, free and online, without the file leaving your device. The columns are found by looking at where the text lines up: a table aligns its columns, and the gap between them is wider than the gap between words, which is a signal prose almost never produces. Charts drawn into the page are located and snapshotted onto a Figures sheet, so they are in the workbook rather than silently lost. That means a tidy table comes across well and a complicated layout needs tidying afterwards. Numbers arrive as numbers rather than text, so you can total them straight away — which is usually the whole reason for wanting a spreadsheet instead of the PDF.

How to use it

1. Add your PDF. It is read in your browser and nothing is uploaded.
2. Each page's text layer is read, and each page is checked for whether it looks tabular.
3. Lines that do look tabular are split into columns on the wider gaps; ordinary paragraphs are kept as single cells.
4. Anything that reads as a number is written as a number, ready to add up.
5. Charts found on the pages are snapshotted onto a separate Figures sheet.
6. Download the .xlsx and open it in Excel, LibreOffice or Google Sheets.

Examples

• A bank statement turned into rows you can total
• A price list from a supplier's PDF, ready to sort

What this tool cannot do

• Columns are detected from where the text sits on the page, because a PDF stores no table structure. Well-aligned tables come back cleanly; ones held together by visual spacing may not.
• Merged cells, nested headers and footnotes attached to a row will not survive as structure.
• Scanned pages go through text recognition first, so check any figure that matters before using it in a calculation.

Frequently asked questions

How does it find the columns?

By where the text lines up. A PDF does not record that anything is a table — it records glyphs at positions — so the structure has to be inferred. The signal that works is spacing: a table aligns its columns, which leaves a wider gap between them than between the words inside a cell. Runs of two or more spaces are treated as column boundaries, and a page is only split that way when most of its lines look tabular, so a paragraph containing a double space is not mangled into columns.

Will my numbers be numbers?

Yes, where they are unambiguous. Anything that reads as a plain number, with or without thousands separators, is written into the cell as a number so Excel will sum it without you converting the column first. Two deliberate exceptions: anything with a leading zero stays text, because turning an invoice reference like 0012 into 12 loses information, and so does anything that is not cleanly numeric. The bias is towards not corrupting your data.

What about charts in the PDF?

They arrive on a Figures sheet as pictures. A chart drawn into a PDF is lines and filled shapes, and the numbers that produced it were used and thrown away — so there is nothing to rebuild a real chart from, in this tool or any other. What we can do is find where the drawing sits and snapshot that part of the page at high resolution, so the chart is in your workbook to look at even though it is not one you can edit. If you need the data behind it, the original spreadsheet is the only place it exists.

Can it read a scanned table?

Yes, with a caveat worth knowing before you start. Pages with no text layer are put through character recognition on your device, so a scanned table does come back. But this tool works out columns from where text sits on the page, and recognition places text less precisely than a real text layer does — so a recognised table is more likely to need tidying, and figures should be checked before you calculate with them.

Is my PDF uploaded?

No. Everything happens in your browser, on your own device. Nothing is sent to us and nothing is stored. Given that the PDFs people usually want as spreadsheets are bank statements, invoices and payroll reports, uploading them to a free web service is a genuinely bad idea — which is why this one does not ask you to.