Onsite Retargeting: A Randomized Field Experiment

We investigate the effectiveness of onsite retargeting, including both a retargeted (product that consumers previously interacted with) and a recommended component (recommendations based on the retargeted product). We use data from a randomized field experiment with an online retailer to compare the effectiveness of two types of onsite retargeting, retargeting based on viewed items (view-based retargeting), and retargeting based on items in consumer’s wishlist (wishlist-based retargeting). Drawing on consumer purchase funnel and preference stability theories, we show that wishlist-based retargeting is more effective in attracting clicks and conversions for the retargeted product; however, its incremental effectiveness declines for the recommended products. We further find that onsite retargeting loses effectiveness as time passes, but this negative timing effect is smaller for wishlist-based retargeting than view-based retargeting. Finally, findings suggest substantial heterogeneity across product and consumer characteristics. This work contributes to personalized recommendations and retargeting literature and provides considerable implications for practitioners.