Rasterizing a PDF page losslessly means every pixel matches exactly what was drawn on that page, nothing more. It doesn't automatically mean the background becomes transparent. Most document pages, whether from Word, Google Docs or a report template, were designed as a solid white page, and the PNG you get back will faithfully reproduce that same solid white background rather than dropping it out.
Transparency only survives if the PDF never had a background
A logo or graphic exported to PDF from design software without a filled background layer can genuinely carry through as a transparent PNG, since there was never any opaque content behind it to render. Pulling a logo out of a full page letterhead or brochure layout won't give you the same clean result, since the page itself was designed with a background. For that kind of extraction, you generally need the graphic isolated on its own page or its own file to begin with.
Watch file size on pages that mix text and photos
PNG compresses flat text and vector content extremely well, but a single page that combines body text with an embedded high-resolution photo, common in brochures or annual reports, drags the whole file size up, since the compression has to handle the busy photographic region too. If one page in a batch comes out unexpectedly large next to its neighbors, a mixed layout like this is usually why.
PNG holds up better when the output is headed for OCR
If the extracted pages will later go through optical character recognition, PNG's lack of compression artifacts around character edges gives OCR software cleaner input than a JPEG, where softened edges around letters can quietly hurt recognition accuracy. For scanned text you plan to keep searchable long term, PNG is the safer archival choice over JPEG.