How to convert a PDF to Excel without retyping it

Bank statements. Invoices. Price lists. Someone sends a PDF full of numbers, you need to total a column, and the only obvious option is typing all of it out again.

You do not have to. But this is one conversion where knowing how it works is genuinely useful, because it is guesswork — good guesswork, and worth checking.

A PDF does not contain a table

When you look at a page and see four tidy columns, what is stored is a few hundred separate pieces of text, each with a position on the page, and nothing at all recording which ones belong together. The lines between your columns may be drawn rectangles. They may not exist — plenty of tables are held together by nothing but alignment.

So every PDF-to-Excel converter, including the paid ones, is inferring where the columns are. The difference between them is how good the guess is.

How the columns get worked out

The obvious method is to group text by where it starts: everything at 72 points from the left is column one, everything at 210 points is column two. It works on a page of labels and falls apart on the thing spreadsheets are mostly made of — a column of right-aligned figures, where 9.50 and 1,204.00 begin in completely different places. Group by left edge and that single column scatters into three or four.

The better approach looks at the empty space instead. Project every piece of text on the page onto a horizontal line, and a column boundary is simply a strip that nothing ever crosses. That works the same for left-aligned labels and right-aligned numbers, because it never asks where text starts — only where there is none.

Converting a PDF to Excel

1. 1Open the PDF to Excel tool. The file stays on your device, which matters more here than almost anywhere — this is usually financial data.
2. 2Drop the PDF in and leave "Recognise numbers" switched on so figures arrive as numbers you can total, not text that looks like numbers.
3. 3If the same table continues across several pages, choose "All pages in one sheet" so one set of column positions is worked out from every page at once and the rows stack up properly.
4. 4Run it, then read the note under the result: it tells you how many columns were found and how many cells were recognised as numbers.
5. 5Open the .xlsx and check it — see the list below for the four checks that catch almost everything.

Four checks before you trust the numbers

• Total a numeric column and compare it with any total printed in the PDF. This single check finds misplaced cells faster than reading every row.
• Compare the row count with the original. A big shortfall means rows were merged, which happens when the source has very tight line spacing.
• Look at the last column. Text that has overflowed from one cell into the next tends to show up there first.
• If a header appears again partway down, the table spans pages — convert again with all pages in one sheet.

Why the dates come through as text

This is deliberate, and it is the one thing people ask about most. Take 05/12/2024. In London that is the fifth of December. In Chicago it is the twelfth of May. A PDF carries nothing that says which convention the document was written in.

A converter that guesses will be silently wrong about half the time, and nobody notices until a report has already gone out. A date sitting in a cell as text is a nuisance you can fix in thirty seconds with Excel's own Text to Columns. A date column that is quietly wrong is a different kind of problem entirely.

When it won't work well

Two situations. A scanned PDF has no text in it, only a picture of a table, so there is nothing to extract — that needs OCR first. And a page that is not really tabular, such as an invoice with a free-form address block, has no column structure to find. If the result comes back as a single wide column, that is the tool telling you the page was prose all along.

Common questions

Can I add up the numbers after converting?

Yes — that is the point. Figures are written as real numbers, so SUM works straight away. Currency and percentages keep their presentation too: £1,234.56 becomes the number 1234.56 formatted as currency, and 12.5% becomes 0.125 formatted as a percentage, so the arithmetic is right and the cell still reads correctly.

Will the columns come out in the right places?

On a normal table, yes. Columns are found from the vertical strips of the page that no text crosses, which handles right-aligned figures as well as left-aligned labels. Where it struggles is a page with no real column structure at all.

Is my bank statement uploaded anywhere?

No. The PDF is read and the spreadsheet built inside your browser tab, which is why it still works offline. Bank statements and payroll tables are exactly the sort of thing that should not be posted to a stranger's server.

Why did everything land in one column?

Because the page has no column structure to find, which almost always means it is prose rather than a table. For a page of text, converting to Word or pulling the plain text out will get you further.

Does it work on a scanned statement?

Not directly. A scan is an image, so there is no text to extract. It needs a text layer adding by OCR first, after which conversion works normally. Try selecting a number in your reader — if you cannot highlight it, it is a picture.