How Do Recommender Systems Affect Sales Diversity? A Cross-Category Investigation via Randomized Field Experiment

Abstract

We investigate the impact of collaborative filtering recommender algorithms (e.g., Amazon's "Customers who bought this item also bought") commonly used in e-commerce on sales diversity. We use data from a randomized field experiment run on a top retailer in North America across 82,290 SKUs and 1,138,238 users. We report four main findings. First, we demonstrate across a wide range of product categories that the use of traditional collaborative filters (or CFs) is associated with a decrease in sales diversity relative to a world without product recommendations. Further, the design of the CF matters. CFs based on purchase data are associated with a greater effect size than those based on product views. Second, the decrease in aggregate sales diversity may not always be accompanied by a corresponding decrease in individual-level consumption diversity. In fact, it is even possible for individual consumption diversity to increase while aggregate sales diversity decreases. Third, co-purchase network analysis shows that recommenders can help individuals explore new products but similar users end up exploring the same kinds of products resulting in the concentration bias at the aggregate level. Fourth and finally, there is a difference between absolute and relative impact on niche items. Specifically, absolute sales and views for niche items in fact increase, but their gains are smaller compared to the gains in views and sales for popular items. Thus, while niche items gain in absolute terms, they lose out in terms of market shares.

A Taxonomy of Online Marketing Methods

“…The cocreation of content is benefitting both the consumers and the corporations. Recommendation systems can help customers to get informed about new products that they like while corporations can enhance their user engagement by providing a personalized shopping experience for the users (Lee & Hosanagar, 2019). Some popular review sites are also using contextual marketing as they target and retarget online users with relevant ads when they leave their webpages (Wu & Bolivar, 2008).…”

Are Review Helpfulness Score and Review Unhelpfulness Score Two Sides of The Same Coin or Different Coins?

“…In that regard, we calculate the Gini coefficient of the votes, based on the vote volume of each voter. Gini coefficient is widely used in the literature to measure the diversity of user behavior [e.g., 22,33]. The Gini coefficient ranges from zero (highest diversity) to one (highest concentration).…”

Facilitating tourists' decision making through open data analyses: A novel recommender system

“…Recommenders as a stimulating marketing technology offer important value to both consumers and firms. Recommenders assist consumers in learning about products/services through large choice sets, whilst at the same time, benefit firms by converting browsers to buyers, promoting cross-selling and increasing loyalty by providing a custom browsing experience (Lee & Hosanagar, 2019). Recommender systems have been developed for various industries to tackle the problem of information overload (Sun, Guo, & Zhu, 2019).…”