Jörn Grahl (University of Cologne):
"How do recommender systems influence consumer behavior? Evidence from a randomized controlled field experiment"
Thursday, November 26, 02:00pm - 04:00pm (WiSo Building, Room 410)
Recommender systems have become essential ingredients of digital platforms. They mine data in order to suggest products, services, or even friends. Recommenders are popular. You would be hard-pressed to find a succesful digital platform without one, but researchers are still struggling to identify causal relationships between recommendations and consumer behavior. Several fundamental questions are not yet fully explored: How do recommender systems influence economic decisions like conversion, sales, or product choice? Do recommenders have a systematic influence on product choice, i.e., do they lead consumers away from established top-sellers toward their own preferences and niches?
To answer these questions, I have designed and implemented the recommender system for the bookseller ebook.nl. The system is based on collaborative filtering (similar to Amazon.com: "Customers who bought this item, also bought:"). We are currently running a large randomized controlled experiment on all visitors of the shop. We randomly assign visitors to a "treatment group" and a "control group". The treatment group is exposed to recommendations whereas the control group is not.
In this presentation I will (a) present the current status of the project, (b) discuss first results, and (c) explain how the project fits into my overall research agenda.