PDF to Excel
Pull tables out of a PDF into a real Excel workbook, reconstructed from column positions.
PDF to Excel
Never uploadedYour file is processed in this tab and never sent anywhere.
How to pdf to excel
- Drop the PDF. Text is read with its position on the page.
- Review the detected grid. Rows and columns are inferred and shown before you download.
- Download the .xlsx. One sheet per page, opening in Excel, Sheets, and Numbers.
About pdf to excel
Financial statements, invoices, price lists, and reports arrive as PDFs constantly, and the data inside them is nearly always needed as a spreadsheet. The problem is that a PDF has no concept of a table. It has text fragments with coordinates. What looks like a grid to you is a set of strings that happen to share an x position down the page, and any converter has to work backwards from that geometry.
This tool reads every text fragment with its coordinates through PDF.js, clusters fragments into rows by their vertical position, then identifies column boundaries by finding the x positions where fragments consistently begin across many rows. Those inferred columns become spreadsheet columns. The output is a genuine .xlsx workbook — Office Open XML, zipped in the browser with fflate — with one sheet per page, and it opens in Excel, Google Sheets, LibreOffice, and Numbers.
Clean, well-ruled tables with consistent column alignment convert well, and that covers most bank statements, invoices, and generated reports. What confuses the detection is anything that breaks column alignment: merged cells spanning several columns, wrapped text pushing a row onto two lines, nested subtables, and figures that are right-aligned in a column whose header is left-aligned. Multi-line rows are the most common failure and usually show up as a row split in two.
The preview exists because of that, and you should look at it. Reconstructing a table from geometry is a heuristic, not a parse, and no tool doing this is exact — including the paid ones. Checking a five-row sample against the PDF takes fifteen seconds and tells you whether the whole extraction is trustworthy. If a scan is what you have rather than a generated PDF, there is no text layer at all, and Image to Excel in the OCR section is the tool that handles that case.
How well different tables convert
| Table type | Result | Watch for |
|---|---|---|
| Bank statements | Usually clean | Multi-line descriptions splitting rows |
| Invoices and receipts | Good | Totals blocks landing in odd columns |
| Generated reports | Very good | Repeated headers on every page |
| Financial statements | Good | Indented subtotals shifting columns |
| Tables with merged cells | Poor | Spans cannot be inferred from position |
| Scanned tables | Empty | No text layer — use Image to Excel |
PDF to Excel FAQ
Why are my columns wrong?
Columns are inferred from where text sits on the page. Merged cells, wrapped text, and mixed alignment break that inference. Check the preview and fix the affected rows in your spreadsheet.
Why is one table row split across two rows?
A cell whose text wrapped onto a second line looks like a new row to a position-based detector. This is the most common issue and is quicker to fix in Excel than to pre-empt.
My PDF is a scan and I got an empty workbook.
A scan has no text layer, so there are no coordinates to cluster. Use Image to Excel, which recognises text in the image first.
Is this a real Excel file?
Yes. It is a proper .xlsx Office Open XML workbook built in your browser, not a CSV with a renamed extension. It opens in Excel, Sheets, LibreOffice, and Numbers.
Can I get one sheet per page?
That is the default. Each PDF page becomes its own worksheet, named by page number.
Is my financial document uploaded anywhere?
No. Extraction and workbook generation both run in this browser tab, which is the point — bank statements should not be uploaded to a stranger's server to be reformatted.