There are two issues. The version submitted to the conference might, at a later stage, be reformatted for publication in proceedings or journals. But both versions should have page numbers.
Page numbers even in a "for review" version help reviewers with any comments back to you and generally ease the work flow. Those page numbers aren't for formal citation since this is pre-publication.
I also think that any published paper longer than a couple of pages should have page numbers to aid readers in making citations to specific parts of your paper.
My experience is that conferences usually want PDF submissions and most people prepare these from a document created by some document editor in which page numbers are easy to add. You may be asked for an edited version later that requires some sort of formatting either with or without page numbers. Follow those directions, of course.
For international submissions, note that page numbering might need to change for final copy. Pagination in 8.5 x 11 won't likely be the same as A4, for example.
I would worry, actually, about conferences that want editable documents for review purposes. If they are, in fact, edited by others, they might change your intent. Journals may need editable documents to allow copy editing and formatting for consistency, but there, the ones doing changes are professionals at that task and, hopefully, get your signify on any changes.