The clickTRUE Blog

Insights on Search, SEO, Social Media, Analytics, Conversions and more.

Predicting The Present: Using Google Trends to Measure Economic Activity – GAAC Summit Day 3

By Adrian Tan on October 19th, 2009

Director, Online Marketing Services. Oversees the SEM, SEO & Web Analytics operations of the unit. Actively engages the industry as a regular speaker; advocating Online Advertising in particular.

in Web Analytics | No Comments

Stock markets are up all around. The Dow Jones just went above 10,000. Home prices back in Singapore are soaring (damn..).  It was appropriate that Google’s Chief Economist, Hal Varian, was here to share with us on how Google Trends could be used to measure and even forecast (or more accurately, “present”-cast) economic trends.

Currently, data from governments usually lag the market. This is normally due to the amount of data processing and verification that needs to be done prior to the release of the information. However, by incorporating trends from Google search query into prediction models, this time lag could be reduced or even eliminated resulting in relevant “present” data rather than lagging data.

As an example, Hal illustrated it with how Initial Claims done by the population (note that these studies are based on the US market) were typically used as a good leading indicator for the end of the recession. There is a considerable time needed for this data to be compiled. To reduce the time lag, we could use Google Trends and obtain patterns of the searches in the Welfare and Unemployment category – by linking the search terms that people would use when researching on Initial Claims. This pattern would be a proxy to that of the Initial Claims. Introducing the variables from Google into the forecasting model, a quick and dirty analysis and forecast of when the recession would end could be produced.

Besides reducing lag time of Government data, Google Trends could also be used to provide a baseline for Intervention Analysis. In other words, predicting how the general population would respond to promotions, incentives and programmes that are in the progress of being announced or implemented. Would the populace defer buying decisions due to an upcoming promotion or programme? What is the impact to the buying cycle when the said promotions and programmes end? This is done by factoring in variables based on search query volume and indicators.

Due to the fact that I am not a citizen of USA (and I do not really keep track of the economic terms in the country, except for being personally vested in the falling US dollar..damn) and that the concepts introduced by Hal were pretty deep in nature, I am pretty sure that I have managed to mangle quite a bit of the concepts presented.

What was important for us was that Google Trends has the potential to introduce much intelligence (and in a very efficient and speedy manner) into either economic models or marketing campaigns (both online and offline). Google Trends could be a very important tool in the years ahead.

For better understanding, you can refer to the two relevant papers on Google’s Research Blog.
Predicting The Present With Google Trends
Predicting Initial Claims for Unemployment Benefits

Also, check out some of the video from the Q&A session (we managed to capture only some segments) between the guys more in tune with economics and Hal Varian.

This concludes our “live” reporting from the GAAC Summit. We will try to put up more articles related to the super concepts we have obtained from this trip once we are back in Singapore. Similarly, we hope to introduce these cutting edge philosophies into our clients’ campaigns down the road.