There are problems with your theory. The city data is portion of searches in that city, not portion of searches overall. So if DC had one search and that one search was "twitter" it would register as 100 even though that is a miniscule portion of all searches. The other problem is that the word twitter is popoular where population is high, which is also where field offices tend to be. Correlation is not causation.
Of course, I know every word I just typed was wasted on you.
Look out behind you, its the FCBIA.