Just to clarify, default conversion preserves Links/Hyperlink annotations, but it does not convert text strings that look like URLs to hyperlinks.
I suppose that you have encountered a file that does not have 'real' hyperlinks. DocPub/PDFNet can extract text and perform pattern matching on text to find potential links with --url_links in DocPub or SetExtractLinks() in PDFNet.