Skip to main content
VellumPDF

PDF to Excel

The tables in your PDFs become usable data again.

Your files never leave your device

Drop your PDF here

or click to browse your files

How to convert a PDF to Excel, step by step

VellumPDF spots tables from where the text sits on the page: a regular grid converts well, a document with no alignment falls back to line by line.

  1. Drop in the PDF

    State a page range to analyze if needed; a password-protected file is refused.

  2. Read the warning shown

    Tables are rebuilt from the gaps that repeat from one line to the next: merged cells, records spanning two lines, or tables without regular spacing may be split incorrectly.

  3. Click “Detect tables”

    If no text could be extracted — a scanned PDF — the tool stops with an error message.

  4. Check the preview page by page

    Each page shows the number of rows and columns detected, or “no table detected” if the blanks are not regular enough.

  5. Tick “Include pages with no table”, if useful

    This option only appears when some pages have a table and others do not; it adds the other pages’ text, line by line.

  6. Choose the format and download

    Excel (.xlsx) or CSV, with a semicolon or comma separator for CSV.

No formula is recreated in the resulting file, only values. On a scanned PDF, run OCR first: unlike the PDF-to-Word tool, the error message here offers no direct link, but the same fix applies.

Frequently asked questions

How does VellumPDF detect tables inside a PDF?

The tool identifies columns from where the text sits on the page, particularly the gutters that repeat from one line to the next. A preview shows the detected rows before download, so you can check the result.

Can I get a CSV file instead of an Excel file?

Yes, you can choose between an Excel workbook (.xlsx), with one sheet per PDF page, or a CSV file using the semicolon Excel expects in Switzerland. Amounts written the Swiss way, such as 1’234.50, are recognized as numbers.

What happens if my PDF doesn’t contain a real table with cleanly aligned rows?

A document without a regular grid can be exported line by line instead of in precise columns. Detection remains local, free and with no file ever sent to a server, whatever the result turns out to be.

What happens if the table I want to extract comes from a scanned PDF?

If the PDF has no selectable text — a scan made of images —, analysis stops with a message saying so: no data can be extracted without a text layer. The document needs to go through text recognition (OCR) first, then come back here once the PDF has been made searchable.

Can a merged cell or a record written across two lines throw off the result?

Yes: columns are recognized when the same alignment repeats across at least 30 % of a page’s lines, with a tolerance of a few points. A merged cell breaks that repetition, and a record spanning two lines of text becomes two separate rows in the export — the preview before download exists precisely to catch this kind of misalignment.

Can I include pages where no table was detected in the export?

When some pages contain a recognized table and others don’t, a checkbox appears to add the latter to the export, taken line by line instead of in columns. Without checking it, only pages with a detected table are exported.