Most likely the PDF page content does not lend itself to export to spreadsheet.
Some variables that impact export (via a service or from Acrobat):
--| Content came from a scanned image.
--| PDF producer used (not all are 'equal').
--| How the PDF producer painted content to the PDF page(s).
--| Source content not mastered as tabular (authoring application lacked an insert table feature or it was not used).
--| Source content provided a complex table via an insert table feature but how this was implemented by the author was not well done.
"Good" export is predicated on have tagged PDF content.
When tagging has to be provided on-the-fly by heuristics it can become dicey.
If the PDF is a well-formed tagged PDF export is, typically, spot on.
A mal-formed tagged PDF can be worse than an untagged PDF.
Not much you can do when all you have is the end of work flow PDF and the PDF is 'off' in some way.
With such you are, basically, in a gigo situation.