My practice is to ignore trees over about 3,000 entries for low or very low confidence matches unless I find a surname or location that I can relate to my family. It is helpful when cousins are added, but because the DNA match surname list contains only the direct line names it's necessary to go into the full tree to find them. That's just too time-consuming for the low probability matches, even though I may miss out on a few possibilities.
I automatically skip any trees that have Julius Caesar and Charlemagne as ancestors.
PS--my Ancestry.com tree has about 7,000 names and it is deliberately exploratory, containing a number of unconfirmed entries. My "official" tree is on my hard drive and includes only verified relatives.