What a PDF actually stores
A PDF does not contain paragraphs. It contains instructions of the form “draw this glyph, in this font, at this coordinate”. Unless the PDF was tagged for accessibility, there is no record that two lines belong to the same sentence, that a run of text is a heading, or that a block of numbers is a table. All of that is something a human infers from the picture.
So a PDF-to-Word converter is almost always reconstructing structure that was thrown away when the PDF was made. This one groups glyphs into lines by their vertical position and lines into a document in reading order, and reads bold and italic from the font name. That is reliable for ordinary prose and unreliable for multi-column layouts, sidebars and tables, where visual order and reading order diverge.
Scanned documents need OCR, and this is not OCR
If your PDF came from a scanner or a phone camera, each page is a photograph. There are no glyphs to read, so there is no text to extract, and the tool will say so rather than hand you an empty document. Recognising characters in an image is a different problem — optical character recognition — and needs a trained model, which this tool does not ship.
To check what text a PDF really holds before converting, run it through PDF to Text: if that comes back empty, the document is a scan. To go the other way, Word to PDF writes real text runs rather than rasterising the page.