1) Kentucky's #1 search term is "free gay porn" - pretty specific relative to all the other terms elsewhere
2) Iowa's #6 is "backroom casting couch" but Wyoming's #3 is "brcc". That's either poor data warehousing practices, or an intentional choice to say "People who type the term in are differentiated from people who use the acronym." I'm assuming that it's poor data warehousing or small sample size given how "milf" is in almost every state but not one state has the fully-spelled out version.
3) Mississippi's results are skewed relative to what most people think about Mississippi. For a state that is supposed to be 60% Caucasian[1], their top search terms are not very Caucasian-like. I'm thinking small sample size here.
4) Connecticut's #10 result is "lisa ann"
5) Maine has both "ebony" and "black" as top search terms
6) Most of the South American listings have some variation of the country name.
7) Every result in Peru is gay. Clearly this is either small sample size, marketing, or "other".
I'm surprised more people aren't focusing on the "dirty data" (pun intended). Clearly something is weird with the data - it's either working on small sample sizes, or there's been some poor work done to get it to this point (not enough work done, for example).