Your file never leaves this browser

Invoice to Excel

Drop a PDF invoice and get its line items as a spreadsheet. The invoice is read here, on your own machine, and nothing is uploaded. Which column holds the quantity, the unit price and the amount is not guessed from the headings: it is worked out from the arithmetic, and then the rows are checked against the subtotal and total printed on the invoice itself. You are told what added up and what did not.

Drop your invoice here

PDF · stays on your device

line items
rows total
header fields
seconds

Questions

How is this different from the tools that advertise 95% accuracy?

An accuracy figure is a probability about invoices in general, not a statement about yours. An invoice, unlike most documents, can be checked against itself: each line’s quantity times its unit price should equal its amount, those amounts should add up to the subtotal, and the subtotal plus tax should equal the total. We run those three checks on your invoice and show you the numbers. If they all hold, the extraction is not likely right — it is arithmetically consistent with what the invoice says. If one fails, you are shown which one.

How does it know which column is which?

From the arithmetic, not the headings. Invoices label the same column Qty, Quantity, Units or nothing at all, so a heading is a weak signal. Instead every combination of numeric columns is tested and the one where quantity times unit price equals the amount across the rows is the real one. On a simple invoice with no quantity column, the amount column alone is read and you are told that is what happened.

What if a row is missed?

That is exactly what the subtotal check is for. If a row is dropped — a page break, a faint scan, a table rule read as text — the remaining rows still multiply out perfectly, so line-level arithmetic alone would look clean. Only comparing the rows against the subtotal printed on the invoice catches it, and that is why an invoice with no readable subtotal is never reported as verified.

Does it work on a scanned invoice?

Yes. A page with no text layer is read with OCR inside this page, so the scan is never uploaded either. The first scanned file is slower because the OCR engine downloads once. Whatever OCR reads still has to satisfy the totals before the result is called verified — a misread digit will not add up.

What lands in the spreadsheet?

The header fields it could read (invoice number, dates, subtotal, tax, total), then a row per line item with description, quantity, unit price and amount, then the result of each arithmetic check. The checks are written into the file on purpose: three months later the file is still there and the screen is not.

Is it free?

The first page converts without an account. Signing in gives five pages a month, and a plan removes the limit. The page count is shared with the bank statement tools, since they are the same engine doing the same work.

Other pdf tools

These all work the same way — in your browser. Most are free; the ones marked as needing a plan are not.