Amazon's business style is unique. While selling a wide variety of products as a retailer, it also provides a digital platform for third-party sellers. On the Amazon website, the same product is often sold by Amazon and third-party sellers. As a result, a competitive relationship arises not only between Amazon and companies engaged in similar platform businesses (such as China's Alibaba) but also between Amazon and third-party sellers who sell products on Amazon.

Any of these points could be made about suggestions from helpful human retailers in small mom-and-pop shops, too. So if there is something to be concerned about in retailing's largest-scale forms of surveillance capitalism, perhaps this isn't it. You don't mention appropriation of consumer surplus through discriminatory pricing, which has been the long-term concern of my former student Professor Tal Zarsky, or anti-competitive conduct with respect to third-party sellers on the platform's marketplace, which has been a focus of regulatory interest by the European Commission, for example. It might be useful to consider what else the information collected by a retail oligopolist like Amazon can be used for, and why the problem of third-party sellers is so significant.

-- By Risako Suzuki - 09 Dec 2021

Introduction

Amazon is the largest retail website as well as the largest platform for third-party sellers, and also provides many other services. I believe that the strongest point of Amazon is the overwhelming ability to collect information. A retail website that treats specific kinds of products can gather information of customers’ preference only relevant to such kind of products. However, Amazon treats a surprising number of and various products as not only a retailer but also a platformer (e.g. food, household articles, electric appliance, books, music, and medicine). Thus, it can gather information on a wide range of customers’ preferences. In addition, Amazon may understand the conversation or life rhythm of customers via Alexa. By using such information and creating an algorithm, Amazon can suggest products customers may be interested in, which has boosted its business.

Data protection law cannot protect customers sufficiently

Under Japanese data protection law (and I assume there is not so much difference between countries), business operators need to specify and publicize the purposes of use of personal data but are not required to obtain consent from customers if they only use them by themselves. If they provide customers’ personal data to a third party, they need to obtain consent from customers. However, as comprehensive prior consent is allowed, business operators usually prepare a privacy policy, list all kinds of information and purpose of use there, and obtain a customer’s comprehensive prior consent on the privacy policy. As for Amazon’s case, when creating an Amazon account, the message “By creating an account, you agree to Amazon’s Conditions of Use and Privacy Notice” is displayed, and by clicking the “Create your Amazon account” button, customers are deemed to have consent on all provisions of the Privacy Notice. As the Privacy Notice is very long, I think almost all customers do not read them seriously. In addition, even if customers read it and consider that they do not want to allow Amazon or its third party to collect and use a certain kind of privacy data, it is impossible for them to agree with some provisions but disagree with the others. In addition, while Amazon provides various services (Amazon.com website, Amazon Music, Amazon Prime Video, Kindle, Amazon Chime, Alexa, etc.), when using any of them, an Amazon account is necessary. So, if a person refuses to agree with the Privacy Notice, he/she cannot use any kind of service provided by Amazon, which means it is virtually inevitable for customers to provide consent on the Privacy Notice and allow Amazon and a third party to collect and use their personal data for any purposes and in any way.

The risk of retargeting advertisement (from a viewpoint other than privacy)

As stated in the first paragraph, I think that, for Amazon, one of the most important purposes of collecting various personal data is to analyze the preference of each customer and recommend a product that a customer may be interested in. The Amazon.com Privacy Notice clearly states as follows in the item titled “For What Purposes Does Amazon Use Your Personal Information?”

Recommendations and personalization. We use your personal information to recommend features, products, and services that might be of interest to you, identify your preferences and personalize your experience with Amazon Services.

However, I think that personalizing the customer experience and recommending a certain product that is chosen based on an algorithm might cause a risk for customers. For example, some products like medicine or food including allergens may harm the health of some people. I assume that the algorithm automatically recommends products based on such as search history or purchase history. Since AI cannot consider the change of health status or food allergy of each customer, there is a risk that it might recommend a product that might harm customers. In Japan, considering such risk, a retargeting advertisement regarding medicine is prohibited (but as for food, there is no regulation). However, Amazon provides its services all over the world and some countries might not have such restrictions.

In addition, recommending a product prevents many customers from searching for products by themselves, which means they are sometimes deprived of an opportunity to purchase a more favorable product. How and based on what information a recommended product is selected is not disclosed to customers. For example, for some customers, price is more important than quality but for others, quality is more important than price, but AI cannot recognize them. If products are selected based on search history or purchase history, even if a new product that is more favorable than the old one appears, it might be difficult for customers to find such a new one. Products sold by Amazon might be more preferentially recommended than those sold by a third-party seller even if the conditions (e.g. price, quality) are not much different. Thus, recommendation and personalization based on customers’ personal information may sometimes disadvantage customers.