Arguably there is no single state more important to beating Trump in 2020 than Wisconsin. J. Miles Coleman on Sabato's Crystal Ball has a useful new article on election trends in Wisconsin 2016-2019. Note particularly the two very interesting maps. My take on what the Democrats will have to deal with and what they must do to win.
To carry the state again, Trump likely needs to increase his support among white noncollege voters from his 19-point advantage in 2016 and/or increase this group’s relative turnout. Alternatively, he could try to increase his support among the considerably less-friendly white college demographic. But the voting patterns from 2016 will likely not suffice for a Trump victory in 2020.
Demographic changes in the underlying eligible electorate would be enough for the Democratic candidate to barely carry the state in 2020, if voting patterns from 2016 remain the same. A safer strategy would be to change some key voting patterns from 2016 in Democrats’ favor. One obvious goal would be to increase Black turnout—which declined a massive 19 points in 2016—back to its 2012 levels. Doing so would add about a point and half to the Democratic margin in 2020.
Widening the Democrats’ already-healthy margin among white college graduates by 10 points would be more effective, adding 3 points to potential Democratic 2020 performance. But moving the Democrats’ white noncollege deficit back to 2012 levels would add 7 points to Democrats’ projected 2020 margin. White noncollege women are the clear target group here, since Clinton’s deficit among these voters (-16 points) was much less than her deficit among their male counterparts (-43 points).
It's doable for Democrats to put together a package of electoral changes that will squeak out a victory. But it's not going to be easy.