PDF to Excel

The problem with every other PDF to Excel converter

They give you a grid of text. It looks right, the columns are in the right places, and then you select the totals column and Excel says zero, because every cell in it is a piece of text that happens to look like a number.

At which point you are retyping it, or fighting text-to-columns and paste-special, and the converter has saved you nothing at all. A spreadsheet whose numbers are not numbers is a table in a document with extra steps.

So the whole job here is deciding what is a number and writing it as one.

How this tool works

A PDF contains no tables. It contains text at positions, and the lines around it, if there are any, are separate drawing instructions that know nothing about the text. So a table has to be recognised rather than read.

The signal that is actually there is alignment. Cells in a column start at the same place on row after row, and ordinary prose never does that. So the text on each line is split into cells at the wide gaps, and any run of consecutive lines sharing those positions is a table. A column is treated as a range rather than a point, because a column of figures is nearly always centred or right aligned, so the same column starts at a different place on every row.

Then every cell is judged. Anything that is genuinely a number is written as one, including the awkward real-world forms: 1,234.56 with its thousands separator, £1,234.56 with its currency symbol, negatives.

And a great deal that looks numeric is left alone

This is the part that takes the care, because each of these is something a real document is full of and each one is destroyed by being helpfully converted.

07:30 is a time. Read as a number it is 7, and a timesheet column becomes nonsense. 2026-08-14 is a date, which a spreadsheet would take as a division.007 is a job number, and turning it into 7 is destruction that looks like tidying. The same goes for phone numbers starting 0800.50% stays as it reads, because writing 0.5 changes what the cell says.84120B and C26-07 are references, not sums.

The preview marks which cells are going in as numbers, so you can see the decision before you save anything.

What you get

A real .xlsx: a sheet for each table, or everything on one sheet if you would rather sort and filter the lot together. Header rows are bold and frozen, so they stay visible down a long table.

It opens in Excel, Google Sheets, LibreOffice and Numbers. There is no watermark, no row limit and no sign up, because there is no server to pay for.

When it will not work

A scan has no text in it, only a picture of text, so there is nothing to find. Run it through recognition first, and then treat the figures with suspicion, because a recognition mistake in a number is much harder to spot than one in a word.

A table with no alignment cannot be recognised as one. If somebody has drawn boxes and placed the text freely inside them, the cells do not line up and there is no grid to find. PDF to Word will still give you the words.

Check it before you rely on it. The structure was worked out rather than read, so a column boundary in an unusual layout can land in the wrong place. That is true of every tool that does this, including the paid ones; the difference is being told.