An interesting Forbes article that was released back in 2010 discusses how Google Search Volume for the term “Jobs” was an effective predictor of U.S. Unemployment rate movements. (See here for the article in question.)
I can imagine that at the time the Forbes article was released that the term “Jobs” was effective at determining changes in the Unemployment rate, but since then, it seems that the track records has not been that great. Looking further into how google trends could be used, I came across Nominal search data for the term Unemployment Benefits and found it to be pretty accurate in predicting the direction of the US unemployment rate. The logic behind choosing the specific term to compare against the unemployment rate is the expectation that someone whom is expected to be out of a job or currently out of a job will have a higher probability of searching the term than someone who is employed and expected to be so for the near future.
The results are quite interesting. At first glance you can tell that the search trend data is not that great at predicting the severity of change in the unemployment rate but it can definitely provide insight into the expected direction with almost a month and a half heads up notice to the Bureau of Labour Statistics release date.