Extract PDF tables to Excel
Upload a PDF with tables and download an Excel sheet with the rows and columns reconstructed.
How it reconstructs the table
Worth being honest up front: a PDF does not store "tables", it stores letters with coordinates. There is no hidden grid waiting to be read; there are loose words, each with a position on the page.
That is why the tool reconstructs the table with a heuristic: it groups words that share the same baseline into a row and, within the row, starts a new column wherever the horizontal gap between two words is wide enough. It works well on aligned tables and degrades gracefully on the ones that are not. Because it is a reconstruction and not an exact read, the result always comes with a warning so you check the columns before trusting them.
The titles you will see are synthetic — "Column 1", "Column 2"… — because the PDF does not say which row is the header. If the first row was the real heading, it lands as the first data row and you can promote it yourself.
It does not run OCR
This matters: the tool reads the text the PDF already contains, it does not recognise images. A scanned PDF — a photo or a document run through a scanner — has no text, only a map of pixels. In that case there is nothing to extract and the tool tells you so rather than returning an empty sheet with no explanation.
The quick test: if you can select the text in the PDF with the mouse, it gets extracted here; if dragging selects the whole image, it is a scan.
Options
- Pages: every page is processed by default. You can narrow it with a range like "1-3,5", which takes pages 1, 2, 3 and 5. Anything outside the document is ignored rather than causing an error.
- One sheet per page: off, all pages stack into a single continuous table; on, each page goes to its own sheet. Useful when each page is a separate table.
- Detect numbers and dates: on, a "1,234.50" is stored as a number and a date as a date, respecting your regional settings. Turn it off if you would rather keep everything as plain text.
Frequently asked questions
Why do the columns come out misaligned?
Because the original table was not cleanly aligned. The tool splits columns at the horizontal gaps; if two columns sit very close or a cell overflows, the split falls in the wrong place. On tables with clear margins this barely happens.
Can I extract only some pages?
Yes. Type the range in the "Pages" option, for example "2-4" or "1,3,7". That is how you skip covers, indexes and pages with no tables.
Does it work on a scanned PDF?
No. A scan is an image with no text and this tool does not run OCR. You would first need optical-recognition software to turn the image into text.
Are the "Column 1, 2…" titles a bug?
No, they are deliberate. The PDF does not mark which row is the header, so neutral and stable titles are used. If your first row was the heading, it lands as a data row and you can use it as such.
What if the result is not perfect?
That is to be expected with some tables: the reconstruction is an approximation, which is why you always see the warning. Fixing a couple of columns is usually far quicker than retyping the whole table by hand.