This case study mainly focuses on Amazon’s Recommendation System (RS) and how the company uses big data within the company’s e-commerce platform. Amazon is a famous American company founded on July 5, 1994, by Jeff Bezos (Hartmans, 2019), and it provides one of the biggest e-commerce platforms in the world. The RS is a machine learning (ML) that uses algorithms to analyze multiple inputs and provide customized results based on specific conditions. A recommendation system is an information filtering system that attempts to predict a user’s rating or preference for a product (Agrawal, 2021). For example, the company personalizes users’ accounts based on various inputs such as items recently viewed history purchases, to name a few, to present products that have a high chance of getting.

Also, Amazon needs an RS to increase the company profits by selling more products, getting more subscribers, and presenting more targeted advertisements by increasing customers’ time on the platform. Amazon’s RS is designed to understand and predict user interests and behaviors and make recommendations based on these insights. It is a critical tool for driving user purchases and increasing and maintaining user attention and engagement on the platform (Singh, 2020).

Meanwhile, this case study shows Spandana Singh’s results in her case study “Why Am I Seeing This? How Video and E-Commerce Platforms Use Recommendation Systems to Shape User Experiences,” published in 2020. Singh represents three recommendation system case studies about how YouTube, Amazon, and Netflix use the RS in their platforms. Also, the writer used other resources to gather information not mentioned in Singh’s case study.

2.   Challenges

2.1. data sources

Amazon has More than one billion terabytes on its servers (Price, 2021). Moreover, it is hard to specify who exactly owns the data within the Amazon company. Also, there is not enough information about the data quality and right to access that Amazon uses to build their RS. Regarding data privacy, Amazon states on their website in the Legal Policies section, “We collect your personal information to provide and continually improve our products and services” (Amazon, 2021). Amazon faces no problems using all data they have in its recommendation system under the conditions of use and any service terms.

2.2. Organizational Challenges

Amazon exists in more than 45 countries worldwide and has more than 12 million products (N, 2022). Additionally, a huge number of third-party sellers buy their products on Amazon.com. According to Arishekar, 140,000 third-party vendors account for more than half of Amazon’s sales (N, 2022). All these factors need high communication to collect the data so the technical department can apply the RS properly.

2.3. Technical Challenges

Building a reliable RS needs to deal with a huge amount of data, real-time data flow, and a variety of data type makes the RS complicated to perform and need high technical expertise and hardware capabilities. Moreover, hence Amazon has different locations, languages, products, and vendors. These different kinds of data and sources also add other technical challenges that need to deal with.

2.3.1.   Data Volume

Amazon is facing big data challenges in different aspects. The company has a large data volume and to ensure the system has accurate results, it needs a lot of data. There are more than 1,000,000,000 terabytes of data across over 1,400,000 servers (Price, 2021), representing the huge amount of data Amazon has.

2.3.2.   Data Variety

Amazon gets different kinds of data from different sources. There are two main sources that Amazon collects the data from, data provided by people and data generated by the platform. Customers, for example, provide personal information, product reviews, and rating products in the cart and Wishlist. Also, there are more than 60 kinds of information that Amazon collects, such as age, location, address, payment, IP address, email, and password, to name a few (Amazon, 2021). Sellers provide product information such as images specifications and answer questions about the products from customers. In addition, there are some data generated by the system while the customers browse the platform, such as pages they visit, products they watch, or mouse movements on screen.

2.3.3.   Data velocity

Furthermore, many purchases happen each minute, increasing the data flow, which requires processing these data in real-time. Also, in the matter of the data flow, Amazon earns $4,722 per second, for $283,000 each minute (N, 2022).

3.   Stakeholders

There are many stakeholders involved in the RS. However, it could be grouped into three stakeholders—first, Amazon company has specific strategic goals to achieve and is responsible for running the RS. Second, the third-party sellers show their products to the Amazon platform to find customers to get their products. Finally, the customers are the most important factor in this cycle. The RS is mainly performed to make the customer buy more products and spend more time on the platform as much as they can to achieve the company goals.

4.   Requirements and Resources needed

There is no information about how Amazon processes its big data operations. Although Amazon has AWS as a service, the writer thinks Amazon uses its software and hardware to run the RS. This indicates that Amazon uses internal recourses since it can use its resources. In addition, as Amazon has the resources to run AWS, they also have the human expertise to perform the system. However, there is no clue about the kind of skills Amazon uses to apply the system.

5.   Time

Amazon introduced a recommendation algorithm to its e-commerce marketplace about two decades ago. It depended on human effort and showed the best-selling products, which introduced some mistakes in the recommendations (Singh, 2020). However, Amazon realized these issues and worked to build solid RS that provided more accurate results. More than a decade ago, Amazon researchers started investigating a wide range of methods to make customer recommendations more useful, including moving beyond collaborative filtering to include individual preferences, learning to time recommendations, and learning to target recommendations to different users of the same account, among many other things (Hardesty, 2022).

6.   Results

Amazon runs the RS to increase its revenues and provide more accurate suggested products. The RS that Amazon builds is successful because it achieved the goals Amazon wanted. Amazon gets from the system $48.0935 Billion. The RS generates 35% of Amazon’s income (MacKenzie et al., 2018). In Dec 2021, Amazon’s revenue at 137.41 billion (MacroTrends, n.d.).

In addition, Amazon also sends recommendations through emails. Such email’s conversion rate and efficiency are ‘very high,’ far outperforming on-site suggestions (Arsenault, n.d.). That was a surprise because the email recommendations are more effective than Amazon thinks.

6.1   learned lessons

Furthermore, Amazon learned from the old recommendation technique, which was inherently biased and did not deliver good suggestions to people with specific interests (Singh, 2020). Using ML and big data offers accurate results, and customers interact with the suggested products, which gains more profits.

Moreover, the RS allows the company to provide a unique personalized shopping experience for every customer. Amazon claims that giving a personalized experience will improve consumers’ overall platform experience (Singh, 2020). Also, Amazon developed multiple recommendation sections that can help customers through their shopping on the platform. There are 13 recommendation sections distributed among the e-commerce platforms (Singh, 2020).

Additionally, because of obtaining the RS, the company gained more revenues and achieved their goals. Also, the customers can find the products they may interested in to buy and see many other products which provides more options. Also, the third-party sellers now can buy more than before because their products have high chance to appear to the right customers.

7.   Critique

The Amazon recommendation system works perfectly, and they use it in many ways. However, there is not enough information about the transparency on how Amazon processes the data to run its recommendation system. Also, the writer thinks if Amazon adds a small survey asking about customer’s preferences, this will enhance the system’s results, and it will be optional to improve the customer experience.

8.   Explain terms

Big data

Big Data refers to the need to parallelize the data handling in data-intensive applications. The characteristics of Big Data that force new architectures are as follows Volume, Velocity, Variety, and Variability (Chang & Grady, 2019).

Cloud Computing Services

It is the distribution of IT resources on-demand through the Internet with pay-as-you-go pricing (Amazon Web Services, n.d.).

E-commerce

Purchasing and selling products or services through the internet, as well as the transmission of money and data to complete these transactions (Shopify, n.d.).

Machine Learning

It is a subfield of artificial intelligence (AI) and computer science that focuses on using data and algorithms to mimic how people learn, progressively improving its accuracy (IBM Cloud Education, 2021).

Machine learning Algorithm:

A procedure that is run on data to create a machine learning “model” and the Algorithms “learn” from data or are “fit” on a dataset. Examples of machine learning algorithms are Linear Regression, Logistic Regression, Decision Tree, Artificial Neural Network, k-Nearest Neighbors, and k-Means (Brownlee, 2020).

Stakeholders

A stakeholder is a person, company, or organization that is interested in or influenced by a company’s activities and the outcomes of those actions (Renfro, 2022).

Streaming Services

An online entertainment provider sends material to a subscriber’s PC, TV, or mobile device via an Internet connection (PCMag, n.d.).

Targeted Advertisements

 It is a method for marketers to provide advertisements to consumers that are tailored to their unique characteristics, interests, and purchasing habits (Lau, 2021).