A PDF does not contain a table. It contains ink in positions, some of which happens to line up. Every converter that hands you back a Word table has guessed, and the quality of that guess is the whole story of this conversion.
What actually breaks, in order
- Borderless tables. If the original has no rules, there is nothing for the converter to detect an edge from, and the cells arrive as tab-separated paragraphs.
- Merged cells. A cell spanning three columns reads as one wide cell in one row and three narrow ones in the next. Most converters split the difference and get both wrong.
- Numbers that were right-aligned. Alignment is positional in a PDF, so a right-aligned column becomes a left-aligned one with a lot of leading spaces.
- Fonts you do not have. Word substitutes, the substitute is wider, and a table that fitted the page no longer does.
Fix it before, not after
If you have any access to the file that made the PDF — the original Word document, the spreadsheet, the export step in whatever produced it — go back one step and export from there instead. Ten seconds of looking beats an hour of rebuilding a table by hand.
If the PDF is all you have, convert it and then check the four items above in that order. They account for almost everything people write to us about.
Worth knowing
A PDF made by scanning is a different job entirely: there is no text in it at all, only a photograph of text. That needs recognition, not conversion, and the result is always a draft rather than a document.
When to give up and send a PDF
If the layout matters more than the editing — a form, an invoice, a certificate, anything with a signature block — the honest answer is that you do not want Word at all. You want the PDF, and you want whoever asked for a Word file to tell you which part they actually need to change.
Most requests for “the Word version” are really requests to change one sentence.