GalleyCat FishbowlNY FishbowlDC UnBeige MediaJobsDaily SocialTimes AllFacebook AllTwitter LostRemote TVNewser TVSpy AgencySpy PRNewser

Tool of the Day: Google Correlate

When it comes to search, Google reigns supreme. Millions of searches are conducted on a daily basis, and that sort of data is valuable to marketers, business people, and journalists as well. We’ve all seen the annual Google Zeitgeist data visualizations which show lists like the most used search queries and the fastest rising celebrities. Earlier this year, Google released Google Correlate, a tool that can mine similar patterns in search data terms.


As you can see, Google Correlate allows you to compare search patterns against a specific time series (weekly or monthly) or against US states. You can use the search box to find terms with a specific pattern of activity or state-by-state correlation. As of this article, Google Correlate does not support international search data, but you can upload your own data set and map it against Google’s search terms. Time-based searches range from January 2003 to the present, and search data is updated on a weekly basis.

The charts are automated using the Google Chart API, and you can zoom in on any particular date range for more information. The units on the y-axis of the graph are standard deviations above mean, so graphs with higher peaks indicate much higher search activity around a specific keyword phrase. You can also toggle your chart display between a line chart and a scatter plot. Perhaps the only drawback here is that your chart can’t be embedded in another HTML document, and the sharing options are limited to Twitter, Facebook, and Google Plus (Google Buzz and Note in Reader share options are available, but both sharing options have been discontinued).

Perhaps the coolest feature in Google Correlate is the ability to approximate real data points from a dataset based on a drawing. Once you draw your curve, Google Correlate attempts to find datasets which map closely to it; the results are ranked by the Pearson product-moment correlation coefficient (r). The higher your value of r, the better the data set maps to your drawn curve. I can imagine how this sort of chart data could be an interesting beginning for more research for a story. For example, why was there such a dip in searches for quotes from “Full Metal Jacket” between 2007 and 2009? Inquiring minds would like to know!

Like all Google Labs projects, Google Correlate is in beta, so new features may be added in the near future. The tool comes with ample documentation in the form of a tutorial, a whitepaper, and even a comic book!

Give Google Correlate a try and let us know what you think. But remember, correlation is not causation.

Mediabistro Course

The Art of the Book Review

The Art of the Book ReviewStarting August 4, get paid to write reviews that will influence the publishing landscape! Taught by a Publishers Weekly book critic, you'll learn how to recommend a book to its audience, write reviews of varying lengths, tailor a review to a specific publication and more! You'll leave this course with two original reviews and a list of paying markets for book reviews. Register now!