How to convert a PDF table to a spreadsheet
- Drop the PDF. The words and their positions are read in the browser.
- Keep the format on spreadsheet. Word, HTML, EPUB and Markdown are in the same picker.
- Download the .csv and open it. Excel, Numbers, LibreOffice and Google Sheets all take it as-is.
How a table gets out of a PDF
The mechanism explains where the accuracy comes from and where it goes. A PDF contains instructions to draw words at coordinates, and possibly some lines that happen to look like a grid; nothing in the file marks a cell as a cell.
So the reconstruction is geometric: words on the same baseline are a row, and the left edges of cells across the whole page are clustered into columns. The clustering is what makes a row with an empty cell stay aligned with its neighbours instead of shifting everything one column left, which is the usual failure of naive converters.
Where it works and where it doesn't
It works on the tables people actually need to get out: statements, invoices, price lists and exported reports, anything where the columns are reasonably aligned. It struggles with merged cells, with text that wraps inside a cell (each line becomes its own row), and with documents that aren't tables at all, which come out as a single column of sentences.
If you only want the words, extract the text instead. For an editable document rather than data, use PDF to Word.
Common uses for PDF to Excel
- Bank statements: get the transactions into a spreadsheet to total them up.
- Invoices and price lists: turn line items into rows you can sort and filter.
- Exported reports: recover the numbers from a report that came as a PDF.
- Research data: lift a published table into something you can chart.
Frequently asked questions
Does this produce an .xlsx file?
It produces a .csv, which Excel, Numbers, LibreOffice and Google Sheets all open directly, and which you can save as .xlsx in one click from there. CSV is also the format you want if the numbers are going into anything other than a spreadsheet.
How does it know where the columns are?
From where the text sits. A PDF has no idea it contains a table; it has words at coordinates. The tool groups words sharing a baseline into a row, then clusters the left edges across the whole page into columns, so a row with a blank cell still lines up with the rows around it.
How accurate is it?
Good on real tables with reasonably aligned columns, which covers most bank statements, invoices and exported reports. Poor on merged cells, wrapped text inside cells, and free-flowing prose, which will come out as one long column. Check the result; the failure mode is visible immediately rather than subtle.
My PDF is a scan.
Then there is no text to place in columns. Run OCR first. Accuracy on a scanned table will be lower, because the recognised word positions are less exact than the ones an exporter wrote.
Is my document uploaded?
No. This matters here as much as anywhere: the documents people convert to spreadsheets are usually financial. Nothing leaves the tab.