The predictive power of search

Updated on:

I was looking through the table of contents of the Proceedings of the National Academy of Science (PNAS) earlier this week, when I ran across this article about how search volume can be used to "predict the present" as well as the near future. The basic idea is that across many different areas, the number of people that perform a Google (or in this case, Yahoo) search for a particular term can be predictive of what those (or other) people are thinking at the time, and may do in the near future. The most famous example of this is probably related to a demonstration last year that search volume could accurately identify flu outbreaks in real time (as people who get flu-like symptoms start searching about the flu online).

The PNAS article is behind a pay wall, but the authors of the paper, who are affiliated with Yahoo, also wrote this blog post about their research (and conveniently left a link to the full paper at the end of the post). In this paper, the authors extend the previous work about the flu to demonstrate that search volumes -- either on their own, or in conjunction with publicly available baseline data -- can be used to predict the success of movie openings, video game releases, and the Billboard charts with fairly high accuracy.

The main point that the authors make is that search volume data isn't always better than other sources of data for many cases, but in so doing, they also show how powerful this type of data really can be (even if other types of data have similar power). We talk a lot on this site about how to optimize your site for search indexing, and about how to use analytics software to track the visitors to your site. Even though the search volume modeling in this paper isn't of direct utility to individual businesses, I think it is a really nice demonstration of the amount of information that can be contained in even limited resolution search data.

The keywords that people search for are strong indicators of their current, and possibly future, state of mind, as is evident from these successful modeling results. As such, the keywords that people use to find your site online tell you an awful lot about those people. Analytics is all about trying to take that information and turn it into meaningful results. Studies like these serve as reminders of how much information there really is to be gained.


Sign up to receive updates in your inbox

We'll send you about two emails per month with tips on how to optimize your LACRM account, and grow your small business. Be the first to hear about product updates, and beta testing opportunities!