Every day hundreds of millions of people type their thoughts, needs, and desires into Google’s search box. Aggregate those and track them over time and you can begin to see patterns, in flu epidemics, the economy, products, candidates — just about anything really.
Once you know what a ripple looks like, and the content of that ripple, you can track it. And you start to see the others. And eventually, you start to identify the ripples that preceded a discrete event instead of the ones that followed. (Via OccasRazr)
Item 1: Google Search Trends and Insights allow you to enter any search term and see how it has relatively trended over the last 5 years. Beanie babies (why beanie babies? why NOT beanie babies!?) for example, had been on the steady decline [thankfully] until recently when they got a huge spike when Teenie Beanie Babies were brought back (whatever that means). The highest search volume coming from Reston, VA. Of mild interest unless you are a seller or manufacturer of beanie babies, then it becomes massively important. If I were looking to start a beanie baby store, Reston, VA would be at the top of my list for potential locations.
Large companies would pay handsomely for an up to the minute trending of their relevant keywords. Imagine being able to have a week or two of advance notice that a product is about to hit it big? I would guess the evidence in searches would predate the data on the ground by weeks or months.
And, what about as a measure of the success of a branding campaign? Companies could track exactly how much of an effect a campaign is having. Are people searching for the brand more? Has one campaign been more successful than others? Did one particular brand exposure have a disproportionate effect?
Item 2: Google Finance Domestic Trends Indexes These track search terms related to various industries. Combined with data on consumer price indexes and consumer confidence, one might be able to predict a rebound in a particular industry. After all, financial markets are built on the confidence and sentiment of the universe of investors and if you can wedge yourself in the time between the upswell of negative/positive feeling about an industry and the increase/decrease of an index, you can rule the world.
(Here’s a tip: the lowest volume of searches around airlines and travel have occurred at about December 15th each year over the last 5. I’m guessing if you want to reserve a room or a plane ticket, that is the time to do it. That also happens to be a good time to buy a car.)
Item 3: Google Flu Trends. Google has “found a close relationship between how many people search for flu-related topics and how many people actually have flu symptoms” and can predict flu outbreaks 1 to 2 weeks earlier than the CDC and is 97% to 98% accurate. (This becomes less useful when cases are in the hundreds as opposed to thousands, becoming far more valuable in an epi- or pandemic.)
As Hal Varian and Hyunyoung Choi (Google’s Chief economist and Decision Support Engineering Analyst) pointed out they are currently only focusing on predicting the present, but given the mounting data available, the diverse channels and types of data available, and the identification of significantly correlated variables in models they will be able to predict the future. (Here is a model that uses Google trend data to improve predictions of changes in credit levels.)
Search can be an early warning of people’s shifting interests and intentions, but it doesn’t tell the whole story. However, the accuracy and importance of these predictions will be ever increasing given:
- the increasing number of people online
- improvements in sentiment analysis and scoring
- increasing importance of real time search (i.e. Twitter search)
- their ability to bring in data from other services such as Gmail, blogs, and news sources