How often Is Ancestry.com's "Thru-Lines" accurate in "Potential Ancestors"?

Not often, I am finding. Often, people are identified as potential ancestors on the basis of their being on many family trees. The problem is that people copy old family trees in trying to grow their own trees. Example: Samuel Porter Glenn 1746-1813 is put forward as 5 Great Grandfather--DNA "SHARED MATCH" "A KNOWN CONNECTION"! Oh, how many family trees he is in! WHAT COULD BE STRONGER? But I can't see one document showing his connection to the purported son, John Glenn, my 4th Great Grandfather one way and 5th Great Grandfather another way. (We believe in keeping the family bloodlines pure.) So I go to the Glenn family DNA and find that my Glenns are of the Renfrewshire tribe (and the Astronaut-Senator is, too), but Samuel Porter Glenn is of another bunch of Glenns. I go with DNA, in the Glenn site, not with SHARED MATCH in Amazon.com, until I see some better evidence than is there now.