PDF to Excel

No, you get the document. The workbook opens on a Contents sheet listing every table with the page it came from, how many rows it has and how many columns, so you can find the one you want without opening twelve tabs. Then one sheet per table, named after the heading the document gave it — an electricity bill comes back with tabs reading the names of its own sections rather than Table 1, Table 2, Table 3. Then a Text sheet with everything that was not in a table, in reading order, with the page beside each line. That last one matters more than it sounds: on an invoice the amount actually being charged is a line of prose above the tables, never a row inside one, so a converter that keeps only the grids throws away the number the document is about. Untick it if you only want the tables.

Because only a table the document actually drew is claimed. A grid needs ruled lines, or three consecutive rows whose text lines up into the same columns with a clear gap between each one. That rules out a lot of real tables — a borderless one with two rows will not be found — and it is a deliberate trade. Text lines up for many reasons: a contents page, an address block, a poem, a code listing. Turning one of those into a spreadsheet is a much louder failure than saying there was nothing there, because a wrong table looks finished. If your pdf table to excel conversion finds nothing, the page is usually text rather than a grid, and our PDF to Word converter will give you something you can work with.

As numbers, and this is the whole job. £1,234.50 becomes 1234.5 shown as money; 45% becomes 0.45 shown as a percentage; (1,234.50) and 1,234.50- both become -1234.5, because those are how a ledger writes a negative. The hard part is the decimal point: 1.234 is one and a bit in London and one thousand two hundred and thirty-four in Berlin, and nothing in a single cell says which. So the question is settled per table rather than per cell — somewhere in a table of figures there is almost always one with two decimal places that only works one way round, and that one decides the rest.

They stay as text, on purpose. A date like 12/05/2026 does not say which number is the month, and a spreadsheet full of dates three weeks out is worse than one full of text. A code beginning with a zero — 007, or a postcode — would lose the zero the moment it became a number. Anything over fifteen digits, like a card or account number, would lose its last digits to the way spreadsheets store numbers. And a spaced number is left alone unless it is grouped in threes, so a phone number does not turn into five million. Each of those is information that cannot be got back once it is gone.

Not by default. A table that carries on past a page break, with the same columns in the same places on the next page, is treated as one table, and the heading row is dropped where the document repeated it. A bank statement across twelve pages is one table to the person reading it, and twelve sheets of four rows each is not something anybody can work with. It is still a guess, though — two unrelated tables of the same shape on consecutive pages would be joined — so it is a tick box you can turn off rather than a rule.

No. A scan is a photograph of a page: there are no characters in the file to read, so there is nothing to put in a cell. You will get a message saying so rather than an empty workbook. Run it through our image to text tool first, which reads the words off the picture, and convert the result. A PDF that came out of Word, Excel, an accounting package or a bank will have a real text layer and is the case this works well on.

No. Choosing a PDF here hands it to code running in this tab, and the pdf to xlsx conversion happens on your own machine — the document, the figures in it and the workbook that comes out never travel. That matters more than usual for this tool, because the documents people want as spreadsheets are invoices, payslips, statements and price lists. There is also an API for the same conversion if you want it from a script, and that one does send the file to our server; the page does not.