Search, Click ... Buy? Using Search to Predict Online Purchases

Study links search data to brand perceptions.

WHEN A CONSUMER TYPES a brand name into Google’s search bar, how often does that specific search lead to a sale? A recent study could help marketers use online search data to predict future sales—as well as to determine the health of their brands.

Among the study's five co-authors are two academics: Jeffrey Dotson, associ­ate professor of marketing and global supply chain at the Marriott School of Management at Brigham Young Univer­sity in Provo, Utah; and Elea McDonnell Feit, assistant professor of marketing at the LeBow College of Business at Drexel University in Philadelphia, Pennsyl­vania. The remaining co-authors are all individuals from Google, including quantitative marketing manager Ruixue Rachel Fan, software engineer Jeffrey Oldham, and advertising research man­ager Yi-Hsin Yeh.

For the study, 1,500 Google users volunteered to take brand surveys about the smartphone and automobile industries, and each gave their consent to have their searches tracked over an eight-week period. Researchers then com­pared the data from these searches to the informa­tion in the surveys to see if users' online search patterns were linked to their perceptions of different brands.

From that data, the researchers iden­tified three primary categories of people doing brand-based searches—those who want to purchase a product, those who have questions about or need to trou­bleshoot a product, and those who are simply interested in the brand.

This categorical context is cru­cial information for marketers, says Dotson. “For example, if someone searches for ‘iPhone problems,’ the word ‘problems’ is going to give you an insight into the motivation for the search,” he says. “You can also look at where they click next. Do they go to Apple Support or do to they go to the Apple Store?”

By identifying the context for each search, marketers can better determine whether users recognize the brand’s name, have positive associations with that brand, are in the market for a prod­uct, or are ready to make a purchase.

It took more than a year for Dot-son and Feit to obtain approval from Google representatives to complete this project, because the company first required the researchers to develop systems that would allow Google users to opt in as study participants and that would protect their privacy once the data collection was underway. Once those systems were in place, “Elea and I never touched the data,” says Dotson. “Our co-authors from Google did all of the data-crunching. It was a very productive relationship.”

“Brand Attitudes and Search Engine Queries” was recently published in the February 2017 issue of the Journal of Interactive Marketing. It also is available at

This article appeared in print in the March/April 2017 issue.