Getting a usable spreadsheet out of a PDF
A PDF has no idea it contains a table. It stores pieces of text, each with a position on the page — and what your eye reads as rows and columns is really just alignment. Converting a PDF to Excel means reconstructing that structure from geometry: grouping text that sits on the same line, then finding the vertical boundaries where one column ends and the next begins. This page does that work locally, shows you the result, and only then builds the file.
Why column detection is the hard part
Columns announce themselves by repetition. A left-aligned column puts every value at the same starting position; a column of figures is usually right-aligned, so the ends line up while the starts scatter. Looking at only one of those edges gets roughly half of real documents wrong. The detection here reads both, keeps the boundaries that repeat across many rows, and discards coincidences — a heading that happens to align with something below it is not a column.
Headers, footers and cells that wrap
Real documents are not just a grid. There is a company name at the top, a disclaimer at the bottom, and descriptions long enough to wrap onto a second line. Treated naively, each of those becomes its own row and the spreadsheet arrives full of debris. Lines holding a single value are therefore treated as continuations: either part of the cell above, or page furniture to drop. That is also why the preview matters — you can see what survived before committing.
What this cannot do
If the PDF came from a scanner, every page is a photograph. There is no text to position and nothing to reconstruct, so nothing comes out. Making that work needs OCR, which means downloading a recognition model tens of megabytes large and accepting a different class of errors. It is not included here, and the tool will tell you plainly when it finds no text instead of handing you an empty file.
Why it runs in your browser
The documents with tables worth extracting are the sensitive ones: bank statements, invoices, payrolls, price lists. Uploading those to an unknown converter hands a copy to a server you know nothing about. Here the PDF is read by PDF.js — the same open-source engine Firefox uses — running on this page, and the spreadsheet is assembled in memory. The only network request is for the libraries themselves.
How to get a cleaner result
A few habits make the difference between a spreadsheet you can use and one you have to rebuild. If the document has several tables with different shapes, extract it once and keep only the pages that matter — mixed layouts confuse any column detection, commercial ones included. If a table continues across pages, choose a single sheet so the rows stay in one block instead of restarting on every tab. If the PDF came from a spreadsheet in the first place, ask whoever sent it for the original: no extraction is as faithful as the file that generated the document. And always read the preview before downloading — it exists so a bad read costs you ten seconds, not an afternoon of checking numbers.
Want to go deeper? Read our guide: How to turn a PDF bank statement into an Excel spreadsheet