On this meta, we have [question-answer-ban] as a tag. It is not used consistently (I just added it to this post), but gives you a quick overview of how others have encountered this, and how they have dealt with it.
The good news is that the ban is based on a metric imposed algorithmically. This means there is no human subjectivity involved in making the call. Of course, the call is based on human interactions (well, I hope all of our users are human and not AIs), which are subjective and unpredictable.
The bad news is that the details of the ban metric are undisclosed. However, I think the human interactions on the site give you a sense how to improve questions and answers.