More Marketers to Turn to Online Search Data for Trend Projections
Marketers have long sought to gain an edge on the competition by spotting trends early. Some researchers have been have been using mainstream media to uncover trends while others have tracking keywords used in search and social media. Academic researchers are now developing methodologies to glean big data from search queries to quantitatively spot new trends and develop projections.
Rex Yuxing Du, University of Houston and Wagner A. Kamakura, Duke University, developed a structural dynamic factor analysis model to incorporate data streams of commonly used search terms for new vehicle sales. To test their model, these researchers compiled search queries used in Google, BlogPulse by Nielsen and Trendistic for Twitter. They gathered data from a 7‑year period and weighted 7 consumer search factors, or data streams. Examples of search streams included words and phrases for foreign brands such as Honda or Nissan, U.S. brands such as Chevrolet or Ford, and European luxury brands such as Porsche and BMW. Not surprisingly, trends in search behavior were found to retroactively correlate to 74% of the variation in new vehicle sales. This held true especially when significant events were studied such as the old Cash for Clunkers program. But correlating other search phrases and terms such as an increase in unemployment was shown to lead to a measurable decrease in the number of searches done on foreign brands such a Honda or Nissan one month later. Similarly, higher gas prices lead to more interest in foreign cars and less interest in brands like Ford two months out.
These researchers suggest that marketers will be able to apply models like theirs to understand which factors most affect future demands for specific products and they will be able to predict the potential rise and fall of sales. This type of information would also allow businesses to confidently develop marketing campaigns to offset the effect of negative trends.[Source: Du, Rex. University of Houston. Kamakura, Wagner. Duke University. Quantitative Trendspotting 2012. Web. 22 Aug. 2012]