As I mentioned in response to "Too Many Groups," most of my greatest progress in research has come as a result of focusing on location more than surnames in general. This approach has helped me make connections that I really am sure I couldn't have made otherwise. It has also enriched my appreciation for different communities in the United States and Canada, and what makes them unique. I've learned more about their geographies, histories, cultures, etc. It has made genealogy more meaningful to me, and made certain areas of the continent more meaningful as well.
The only group I've started so far has been for a specific, unique geographic region. I felt a desire for that area to have a group because of the way it differs from the rest of the state it is in, and how many of my own extended "family" members I have found there, in the handful of counties contained.
County groups work best for some research, state groups for others. Maybe even city/township groups work. At the nation level, it can be less effective unless one has specific information or recent immigrants in the family, or both.
I have a particular fondness for multi-county regions, myself. It has worked really well for me. I'm sure one of the reasons why is that several branches of my family got here long enough ago that there were fewer, larger counties. As the populations grew over decades, they had to be split up for proper governmental representation/administration. Thus the families that lived in the original district were now separated by new county lines. Sometimes, they'd move between pieces of property in 2 or 3 different counties, and back. I've seen it happen with families in South Carolina, Louisiana, Kentucky, Illinois, New York, Michigan, and Ontario in particular, in my ancestors.
So - think regionally!