Some number of people just smack the "spam" button for nearly everything that is automated, and those "spam" buttons seem to work on absolute numbers not percentages; so if you have a high number of people in the pool then you will be false flagged eventually.
We had a very explicit double opt-in system, made it super easy to unsubscribe, emailed once a month at most; and we had people still marking our communications as spam. I'm not sure what else we could have done to weed out the people who just smack the spam button honestly.
That said, there was a lot of variance, emoji in the headline was the campaign that caused 9% of people to mark spam and 20% to unsubscribe, but it was enough to have us blackholed for 2.5 months.
I think a major issue is that people don't want to even check how to unsubscribe and they see the “mark as spam” button as a “just make this go away” button.