PDF to CSV: what is really going on
A PDF is a description of where ink goes on a page, not a table. Converting it to CSV means reconstructing structure that was never stored. That is why the same tool can be excellent on one file and useless on the next.
Two kinds of PDF
Digital PDFs contain a text layer — the words are selectable. Scanned PDFs contain only an image. The first converts near-perfectly and instantly; the second needs OCR and will make mistakes. Knowing which one you have explains almost every conversion result you will ever get.
Why the columns drift
Text in a PDF is stored as fragments with coordinates. Nothing says which fragments form a row. Tools guess using vertical alignment, and they guess wrong on merged cells, right-aligned amounts sitting far from their labels, and two-column layouts. That is the failure you see as 'the numbers ended up in the wrong column'.
When to stop converting and start extracting
If the document is a form — an invoice, a receipt, a statement — do not try to rebuild the whole table. Extract the four or five fields you need and let a human fill the rest. Full table reconstruction only pays off for genuinely tabular PDFs with a stable grid.
Checking the output in thirty seconds
Sort by the amount column. Check the largest and smallest values against the document. Check the row count against the page count. Those three checks catch nearly every structural error.
Common questions
Is PDF to CSV lossless?
Never. Even a perfect conversion loses layout, fonts and the relationship between visually grouped items.
Can I convert a password-protected PDF?
Usually you must remove the password first; most browser tools cannot open encrypted files.
What about tables spanning pages?
Each new page loses the header, so the rows land without context. Handle page by page and repair the joins afterwards.