It is a nicely designed chart – although according to discussion on their forums and even my own results, the specificity seems to affected by the current size of the comparison data set.
My ancestry is 50% Irish and 50% Czech and I can name the exact towns where 6 out of 8 of my great-grandparents came from. My 99.6% reported European ancestry is purported to be in the “sweet spot” of their data – however my results have a fair amount of “Nonspecific” percentages. They also show what I think is probably noise in their analysis with tiny amounts of South Asian, Middle Eastern and North African in there as well.
With 23andme’s big push to reach a database of a million people, it will be interesting to see how this composition chart changes over time – assuming they update it regularly as new information becomes available. (Some other reference info on the site has not seemed to keep up with the times.)
Overall, I am quite happy with the service and hopefully I’ll find a match one day that will help me find where them long lost Tierneys came from back in Ireland.