My wife's been doing genealogical research for the past 10 years or so. She takes the quality of her research seriously, often requiring multiple corroborating sources before she's willing to consider a connection "confirmed".
She sees DNA testing as a very handy tool, but you have to realize it's limitations. Results have to be viewed in context with other sources of information, and the specificity of matches drops off dramatically after a couple of generations.
She says that it's also a very fast-moving field, with the quality and depth of the test results constantly improving.
The "ancient ancestors" marketing ploy is double-edged. It gets a lot of people into the database, but many of those people read their results, find out that they're 2% Mongolian or Sub-Saharan, and then never log into the genealogical sites again. (Background: These sites provide messaging services so that members can be notified of the existence of people with significantly matching DNA and can "ping" each other anonymously to request further contact.) This has proven very frustrating for the active researchers, who get a notification that there's, say, a second-cousin match with someone whom they now can't contact. SO there's a ton of potentially useful information wasting away in the databases.