Unexpected statistics sometimes work surprisingly well in measuring in a more timely manner than official numbers some important economics aggregates. The prime example is the recession index of The Economist, which is based on the frequency of the word "recession" in the Wall Street Journal and the New York Times. This measure agrees well with the NBER business cycle dates, and does so much earlier than official data is released or the NBER ruled. Of course, one could argue that there is an endogeneity issue with this index, and journalists may talk up a recession, and then it really happens.
Nikos Askitas and Klaus Zimmermann offer another indicator that does not suffer from such endogeneity problem. They claim that the German unemployment rate is appropriately measured by Google searches using some keywords, like (translating) "unemployment office or agency", "unemployment rate", "personnel consultant" or the most popular job sites. What can Google not do?