Skip to main content
EDIT

The Value of Personal Data in Internet Commerce: A High-Stake Field Experiment on Data Regulation Policy

The Value of Personal Data in Internet Commerce: A High-Stake Field Experiment on Data Regulation Policy

In a recent article forthcoming in Management Science, Professor Tianshu Sun and his co-authors explore the impact of strict data privacy laws on e-commerce by performing a large-scale randomized experiment on Alibaba, the largest e-commerce platform in the world.

03.22.23
Color photograph of a mobile phone displaying the AliExpress app in iOS.

Stay Informed + Stay Connected

MARSHALL MONTHLY BRINGS YOU ESSENTIAL NEWS AND EVENTS FROM FACULTY, STUDENTS, AND ALUMNI.

E-commerce represents over 22% of total retail sales and continues to grow rapidly worldwide, accounting for about $3.3 trillion. Most e-commerce platforms provide product recommendations on their home pages tailored to individual users based on their past browsing and purchase behaviors at the e-commerce site as well as other sites. The platforms develop models of the user preferences using this data and recommend products and merchants based on the models. Over the past decade, policymakers worldwide have raised significant concerns about the privacy and proper use of personal data and laws passed restricting the acquisition, storage and use of personal data. However, little is known about the impact of privacy laws on e-commerce.

In a recent article forthcoming in Management Science, Professor TIANSHU SUN and his co-authors explore the impact of strict data privacy laws on e-commerce by performing a large-scale randomized experiment on Alibaba, the largest e-commerce platform in the world. This study tries to understand how recommendation systems that do not use consumers’ personal data impact consumer browsing and purchase behavior. The study involved an experiment on the Alibaba site for about half a million consumers who visited the site. Half of these consumers were assigned to a control group and for this group, the recommendation system used personal data of the user (such as demographics, past browsing, clicking and purchase histories) along with product and merchant data to make product recommendations placed on the home page. The other half were assigned to a treatment group and for them, the user’s personal data was not used in making product recommendations on the home page.

The study found that the click-through rate on the recommended products was a dramatic 75% lower in the treatment group as compared to the control group. Moreover, those in the treatment group reduced their browsing on the home page and reduced their visits to the e-commerce site. This resulted in a whopping 81% decline in purchases ($ value) among those in the treatment group as compared to the control group. The study demonstrates that recommendation systems that do not incorporate users’ personal data in providing product recommendations can result in substantially lower engagement and purchases by users. This is primarily because the appropriate matching of consumer preferences and product/merchant attributes is substantially lower when consumer personal data is unavailable.

The study also found that smaller merchants and products that are unique and rare are less likely to be purchased if recommendation systems do not avail of personal data. Thus, large well-known merchants are not impacted significantly by the restriction on use of personal data. Instead, it is the smaller, niche merchants who benefit the most from such e-commerce platforms that get hurt the most if personal data cannot be used by the recommendation systems. On the customer side, browsing behavior and purchases decline most among those who are newer to the site, female and from smaller cities and towns.

Overall, the study provides important insights into the value of personal data in internet commerce and the impact of data regulation policy on consumer behavior and e-commerce revenue.

"The Value of Personal Data in Internet Commerce: A High-Stake Field Experiment on Data Regulation Policy"

Tianshu Sun, et al.

GO TO PAPER