Content creation is producing what actually plays on the screens. On a signage project it is also the line item that gets estimated once and then recurs forever, which is why networks go stale rather than break.
Screen content is not repurposed print or social. Viewing distance sets the minimum type size, dwell time sets how much can be said (three to five seconds in a corridor, minutes in a waiting room), and the loop length decides how fast a passer-by sees a repeat. Content that ignores those constraints is technically playing and practically invisible.
The caveat is that the honest question is not "who designs this" but "who designs this every week, forever". A launch batch is easy; the failure mode is month seven, when the same six assets are still cycling and staff have stopped looking. Templates that let a non-designer swap a price or a photo are worth more than beautiful bespoke creative nobody can update.
Feeds and generated content solve part of this by making some of the screen change itself, which is a better answer than resolving to try harder.