What this does

This finds the tables in a PDF and writes them into a real Excel workbook — an .xlsx file that opens in Excel, Numbers, LibreOffice and Google Sheets. Figures come across as numbers, so a column of prices can be summed the moment it opens rather than after an hour of retyping.

The whole conversion happens in this page. Your PDF is never uploaded, which matters more here than on most tools: the documents people want as spreadsheets are bank statements, invoices, payroll runs and price lists.

How it finds a table

A PDF doesn't contain tables. It contains characters at coordinates, and — if you're lucky — lines drawn around them. Rebuilding a table from that is inference, and this tool uses two sources of evidence in order of how much they can be trusted.

Ruled lines come first. When a document draws its grid, as invoices, statements and spec sheets nearly always do, the lines say exactly where the rows and columns are and there's nothing to guess. This is the path that produces reliable results.

Alignment is the fallback, and it's off until you turn it on. Text that lines up in consistent bands down the page is read as a table, subject to some scepticism: fewer than three rows isn't a table, columns whose left edges wander by more than a few points aren't columns, and rows that disagree about how many columns there are mean the page is prose. Those rules exist to stop paragraphs being served up as spreadsheets — but they're rules of thumb, and you should look over anything they produce.

Numbers, and why they matter

The difference between a useful spreadsheet and a picture of one is whether you can total a column. A figure printed as £1,234.50 is text until something strips the currency symbol and the thousands separator, so this tool does that, along with the other ways numbers get printed: a trailing percent, a Unicode minus, and accounting's parentheses for negatives, so(1,234.00) arrives as −1234.

It's deliberately strict about it. Anything left over after those known decorations means the text stays text — 12 boxes is not 12, and a part number that looks like a decimal isn't converted into one. If you'd rather nothing was touched, turn offRead figures as numbers and every cell arrives as text.

How to use it

Drop in a PDF.
Leave the defaults unless the tables have no ruled lines — then turn on unruled detection.
Convert, and download the .xlsx.

What it won't do well

Scanned pages have no text to read. If the PDF is a photograph of a document there are no characters, only pixels, and nothing can be extracted. That needs OCR, which this tool doesn't do. It will tell you it found nothing rather than hand you an empty workbook.
Merged cells don't survive as merged cells. A cell spanning three columns lands in the first of them and the other two are left empty.
A table split across pages becomes two tables, one per page, because nothing in the file says they're the same table.
Formatting is not carried across. Colours, borders and fonts are dropped; you get the values and a bold header row. The point is the data.
Formulas never existed. A PDF holds the result a formula produced, not the formula, so what you get is the numbers.

Frequently asked questions

Is my file uploaded?

No. It's read and converted in your browser. You can watch the network panel while it runs, or disconnect from the internet once the page has loaded and convert anyway — it still works.

Why did it find nothing?

Usually one of two reasons: the pages are scans with no text layer, or the tables have no ruled lines and unruled detection is off. Try turning it on. If the page is a scan, no converter can help without OCR.

Why is one column full of text where I expected numbers?

Something in those cells isn't purely a figure — a footnote marker, a unit, a stray space inside the number. The tool errs toward leaving text alone rather than inventing a value.

Can it handle password-protected PDFs?

Not yet. Remove the password in whatever opens the PDF, then convert the unprotected copy.