Online Shopping Searches Go Far, but not Wide

September 24, 2016
Behavioral Science, Marketing

When buying big-ticket items online, consumers tend to spend a lot of time searching but explore only a small slice of the available choices, spending nearly a third of the time looking at items they've already viewed.


 Professor Carl Mela

Those are among the findings of research from Duke University's Fuqua School of Business that studied online camera purchases. It was one of the first to explore the nature of internet search outside the lab by combining the sites consumers visit with the details about the products they view.

"Our goal is to explore search in the wild," Professor Carl Mela said. "Advances in data mean it's now possible to see how search evolves as consumers narrow down their search for a product.

Though as it turns out, search doesn't actually narrow that much. Mela, along with Bart Bronnenberg of Tilburg University and Jun Kim of the Hong Kong University of Science and Technology, found that the quality consumers begin searching for — price, or a particular feature — is usually the one they're still most interested in by the end.

The findings, "Zooming In on Choice: How Do Consumers Search for Cameras Online?" are newly published in the journal Marketing Science.

The team scoured data from almost 2 million consumer visits to three major online retailers — Amazon, Best Buy and WalMart — over three months in 2010, to find people who searched for cameras. They reviewed the complete 3-month browsing histories for those people, and examined the content of the camera product pages they visited.

That sweep netted 967 cameras purchases they could study from beginning to end.

"A great use of big data is that we could cast our net really widely to find the small fraction of customers who are in the market for cameras at any given time," Mela said.

The average process lasted 15 days from initial search to camera purchase and included six browsing sessions. The average buyer conducted 14 searches for specific make and model, five of which were repeat searches.

"Almost a third of search volume is people revisiting what they already searched," Mela said. "Just like there's loyalty in purchases, there's loyalty in search."

Shoppers generally found their purchased model late in the search process, but searched only a few of the choices available.

"They only search a tiny fraction, only 1 or 2 percent of the possible range of options," Mela said.

As search continues and shoppers begin comparing models, the process also speeds up, the researchers found.

"As consumers begin to get a better sense of the options and how they configure, items become more comparable," Mela said. "It's easier to compare like bundles of attributes than different bundles of attributes, so search accelerates."

A clearer understanding of how consumers search for products can make retailers better at helping zero in on what they're looking for more efficiently, Mela, said.

"Advertisers or retailers can learn about people's preferences from their search," Mela said. "By the time a retailer observes purchases, it's too late to influence it. But with search data, one can determine customer interests prior to purchase, and offer better options to consumers. Search is very useful for determining preferences early on."

This story may not be republished without permission from Duke University's Fuqua School of Business. Please contact for additional information.

Contact Info

For more information contact our media relations team at