This paper investigates how Amazon's marketplace may steer customers toward certain "Related Sellers" (sellers who use Amazon's subsidiary services like fulfillment or shipping). Key findings include:
Amazon’s algorithmic offer selection often favors Related Sellers, even when customers' own preferences differ.
Amazon applies different performance evaluation standards for Related Sellers, sometimes showing misleading performance metrics in their favor.
The number of seller ratings influences customer choices the most, but this metric may reflect a seller's size rather than the actual service quality, indirectly favoring larger Related Sellers.
When customers are presented with corrected performance metrics, their preference for Related Sellers drops by nearly half.
The study highlights potential biases in Amazon's seller selection process, which may disadvantage smaller, independent sellers.
This paper examines the impact of interleaving sponsored (advertised) results with organic results on search engine result pages (SERPs), specifically focusing on Amazon's digital marketplace. While ads help businesses reach consumers, they can also negatively affect search quality and competition:
Sponsored ads often push less relevant, more expensive products to the top of search results, leading consumers to potentially worse choices.
In many cases, products that rank poorly (beyond the 100th organic result) appear as top sponsored results on Amazon's first page.
These sponsored items are often more expensive and of lower quality than the top organic results.
The study, based on 4,800 searches across four countries, highlights the need for greater transparency and fairness in advertising practices on digital marketplaces.
This paper compares traditional desktop e-commerce search to voice-activated searches using virtual assistants (VAs) like Alexa. In desktop search, users see a ranked list of products, while in voice search, Alexa usually recommends just one product with an explanation and a default action (adding the item to the cart). The study finds that Alexa's explanations and product choices often differ from users' expectations:
In 81% of cases, users' understanding of a "top result" differs from Alexa's.
For 68% of queries, Alexa suggests products less relevant than those shown in Amazon's desktop search.
In 73% of cases, users prefer the desktop search result over Alexa's suggestion.
These findings raise concerns about fairness and trustworthiness in voice-based e-commerce searches.
This paper audits Amazon's algorithmic recommendations, focusing on concerns that Amazon's marketplace unfairly favors its own "private label" products over competitors. Key points include:
Producers and sellers worry that Amazon's algorithms are biased toward its own products, especially as ad-driven "sponsored" recommendations replace organic ones.
The study proposes a network-based framework to measure and compare biases in organic versus sponsored recommendations.
Findings show that sponsored recommendations are significantly more biased toward Amazon's private label products than organic recommendations.
While relevant to Amazon sellers, the bias measures introduced are broadly applicable to any platform using recommendation algorithms.
The study highlights the need for transparency and fairness in e-commerce recommendation systems.
In digital markets, antitrust law and special regulations aim to ensure that markets remain competitive despite the dominating role that digital platforms play today in everyone's life. Unlike traditional markets, market participant behavior is easily observable in these markets. We present a series of empirical investigations into the extent to which Amazon engages in practices that are typically described as self-preferencing. We discuss how the computer science tools used in this paper can be used in a regulatory environment that is based on algorithmic auditing and requires regulating digital markets at scale.