NEW YORK, NY – Online shopping has increasingly become the primary method for searching for and buying products. During the holiday shopping season, sales for consumer products typically surge, especially for online shopping. With fears that consumer spending will slump this holiday season due to rising inflation and limited consumer savings, it becomes crucial for the biggest brands to find and reach potential customers. New research from Columbia Business School reveals that companies and advertisers are relying on flawed tools to understand customers’ online shopping behaviors. The prevailing assumption among advertisers is that customers actively choose to search for online products. However, a recent study by Columbia Business School Professor Andrey Simonov finds that searches initiated by ads—which drive 53 percent of website traffic—yield significantly different results than searches initiated organically. More specifically, shoppers who click on a website from an ad spend less time on the website, search for fewer items, and are overall less likely to make a purchase.

For the study, titled Online Advertising as Passive Search, the researchers conducted tests on a substantial dataset of customer searches for apparel. They discovered that the current model widely used by advertisers—which doesn't take advertising into account—significantly overestimates customers’ preferences by 18 percent. In contrast, their model is 48% more accurate when analyzing searches for frequently advertised products and 29% more accurate when looking at purchases of frequently advertised products. The study introduces and evaluates a novel model that helps companies better predict and understand customers’ online search patterns and purchasing choices. Professor Andrey Simonov and his co-authors, New York University Professor Raluca Ursu and doctoral student Eunkyung An, developed a straightforward model that differentiates between "active" and "passive" search decisions. In this context, "active" refers to instances where shoppers intentionally search for a product, while "passive" denotes situations where they simply respond to encountering an online advertisement. 

“Our research shows that consumers are not always actively seeking out the best product, and instead, they are simply reacting to ads that come across their screen,” said Professor Andrey Simonov, the Gary Winnick and Martin Granoff Associate Professor of Business. “This finding is important because it may inflate how companies translate customer preferences. Companies should incorporate a more holistic approach that factors in both passive and active search if they want to better target and understand their customer base.”

For this study, Professor Simonov and his co-authors describe and model how consumers search for products when exposed to ads by analyzing a dataset from German market research company GfK with the full internet browsing histories of nearly 8 million “clicks” of over 4,600 customers in the Netherlands who shopped on over 1,000 apparel-related websites between February and May of 2018. Examining the full website URLs included in the dataset, the researchers identified which website clicks were organic or ad-initiated. They then compare their new model—which separates organic and ad-initiated clicks—with the old model, which treats every search as an active decision known as the Weitzman model and show that their new model is much more accurate and relevant for advertisers.

Additional Findings Include:

  • Customers are More Likely to Search Through Ads Early in the Search Process – The researchers find that ad-initiated searches make up 22% of searches early in the search process but only 7% in later phases.
  • Shoppers Rarely Have a Prior Relationship with Brands They Visit Through Ads – Their data suggests that for 90% of cases where someone clicks on an ad and goes to a website, it’s their first time visiting that website.
  • Most Ad-Initiated Searches Happen When Customers Are Online But Not Shopping – Customers frequently click on ads while they are engaged in other online activities such as checking email or visiting social media websites.

“In this day and age, as brands transition to e-commerce and markets have become saturated with floods of customers, companies need the right tools to implement marketing and advertising strategies. Adopting best practices is even more important around peak shopping times like the holidays. Our research presents a model that helps refine this process,” says Professor Simonov.

To learn more about cutting-edge research being conducted, please visit Columbia Business School.


About the Researcher

Andrey Simonov

Andrey Simonov

Gary Winnick and Martin Granoff Associate Professor of Business
Marketing Division