I've found that there are three primary types of geoguessr player.
The educated guesser may travel down the road a bit, but in the end makes a raw guess based on what they see alone. I've seen average scores in the neighborhood of 11,000-20,000. Some of these players also see anyone who does research (extensive or otherwise) as cheaters.
The researcher may travels down many roads and uses the internet to try and get close to the location.
The obsessive compulsive (me, btw) goes to any length and uses google maps w/ street view to try and pinpoint the locations as closely as possible, often within a couple of meters. They often get the max score of 32,395.
I'm generally amused by those that think that doing research is "cheating". The game does not come with any rules about it so the manufacturing of one is curious. I think it's a measure by which the various types of players hold their score relative to others. "If you get a much higher score than I do, but play differently (i.e. not the way I play), then I categorize your method as cheating." Rather than, say, accepting that different types of players are acceptable.
The fact that there is a maximum score suggests that attempting to reach it is a valid game goal.
In the end, there's room for all types of players and the recent inclusion of timed challenges let others play on equal footing.