Sentiment Analysis (SA) of customer reviews on dietary supplements: A study of Amazon.com.


Wittenborg University



Module Code

Sentiment Analysis (SA) of customer reviews on dietary supplements: A study of Amazon.com.


Worldwide events such as SARS-CoV-22 pandemic that started at the beginning of 2020 have caused diverse changes in the routine actions of people around the world. As the pandemic has evolved in early 2020, individuals conscious of their health stocked on dietary supplement products throughout the year (J.P.Morgan, 2020). As virus spread around the world, more people in the US have become more cautious of their well-being and health, and as studies show, they have emerged a habit of buying any dietary supplement product that could potentially improve their current wellbeing. The tendency of staying home during lockdown and making purchases online has become the part of daily life for people around the world, including US. For instance, Amazon.com over recent years has become the top destination for online shoppers in the US. Moreover, Amazon.com reaped gigantic profits amidst pandemic (Harris, 2020).

Customer reviews are known as one of the most powerful E-Commerce tools. Amazon.com is not only beneficial exclusively for its consumers, who are trying to find the ideal product to purchase, but the marketplace could help marketplace sellers to boost sales and retain customers. A report by USA Today states that customer reviews on Amazon.com are ranked highly because of their trustworthiness (Weise, n.d). In order to pertain the power of customer reviews for the purchase of the products, Amazon.com has issued more than a thousand of lawsuits or the fake reviews. Indeed, the report confirms that Amazon.com gives preferences and places the reviews on top of the list of customers who had purchased the product (Weise, n.d). The help of machine learning is included in the process of eliminating the fake reviews to ease the customer experience of buying a product.The power of a customer review who made a purchase on influencing other potential customers on Amazon.com is undeniable, as the word-of-mouth of sharing an experience to ten friends is different than sharing an experience to millions of people. Customer reviews have become a substantial part of purchasing decisions and subsequently, a buying behavior on marketplaces such as Amazon.com, and an analysis undertaken in 2017 revealed that 84 percent of 100 percent of shoppers trust online reviews on Amazon.com as much as personal recommendations of close peers (Inc, 2017). Customer reviews help to create transparency in the buying experience of consumers, who are more likely to make a purchase. Consumer reviews and feedback reveal consumers’ positive or negative experiences in relation to factors such as price, value, customer service, product quality, ease of purchasing, and many more aspects on why customers make online purchases on marketplaces. Indeed, the buying behavior of consumers varies depending on different geographical, social, and cultural aspects. For instance, dietary supplement products such as minerals and vitamins in tablet form may not be widely bought and consumed in third-world countries. In contrast, developed countries such as the US consume dietary supplement products on a regular basis and tend to consume dietary supplement products more often than other nations. The most common reasons adults in the US consume dietary supplement products regularly are to improve their overall health and to maintain health (Bailey, et. al., 2013).

SARS-CoV-19 has also prompted the regular consumption pattern of different dietary supplements such as multivitamins that prevent the widespread of bacteria and viruses. Dietary supplement products are aimed to add or supplement the regular diet of a person and are different from the regular food. Dietary supplement products come in different forms and types, such as tablets, soft gels, gummies, capsules, herbs, powders, bars and liquids (U.S. Food and Drug Administration, 2019). These products are not usually prescribed by the specialist, however are considered and stated as drugs.


As was mentioned, the marketplaces as Amazon.com have become a substantial part of shopping behavior in the US, replacing regular brick-and-mortar shops. As different studies were implemented over the years it is crucial to study and analyze the customer reviews of dietary supplement products such as vitamins and minerals as the reviews of other consumers could potentially serve as the recommendation to purchase the product.


3.   Focus of the study (Problem Statement)

Customer reviews are important in the marketplaces as part of the user-generated content, as anyone has access to write a negative or positive review. Indeed, a bad customer review can damage a reputation and overall perception of a brand, while a fake review could unwillingly bring a wrong perception of a business and a product and most importantly impact the purchasing decision of a customer.A customer review on Amazon.com is a crucial indicator for potential and existing customers. Buyers commonly are more likely to purchase items that other consumers have purchased and claimed to be good. The uncertainty of consumer reviews’ fairness and a low number of sales could be a challenge to attracting new customers. Indeed, there is a strong need for marketplace companies to reach out to consumers more effectively and compellingly.

Dietary supplement products are not allowed to be marketed as a treatment, cure or prevention for any disease, however, sales of these products have emerged after COVID-19 (Hamulka et al.,2020). Moreover, many people were hoping that dietary supplements might help to prevent the virus. Apart from that, the study implemented by Mordor Intelligence states that the sales of dietary supplement products in the US in  the last few years is explained by people’s high stress levels, hectic routines, which has led to unavoidable lifestyle diseases such as obesity, diabetes and high blood pressure (Mordor Intelligence.,n.d).Other than that the study analyses that customers in the US tend to purchase and use more dietary supplement products because of their diverse application. For instance, nutritional supplements are used by individuals who suffer from heart diseases, orthopedic issues and for sports. Nevertheless, it is assumed that the trend of following the healthy lifestyle influenced by celebrities and social media channels have strong impact on purchasing supplements as gummies, powders, and tablets.

The problem that will be studied in this research is how dietary supplements are reviewed on Amazon.com by customers and important factors marketplace sellers need to consider to increase number of sales. By determining the important components of established purchasing behavior as a result of purchasing dietary supplement products, this study will strive to give recommendations for further research in this topic. Moreover, business insight derived from the results of the analysis performed will be provided for existing and new Amazon.com sellers. The significance of the research will impact on the factors customers focus on writing a review on a product on Amazon.com.

There were numerous studies implemented on Sentiment Analysis (SA) of customer reviews on Amazon in different countries, however there is a gap in studies that analyses the reviews on dietary supplement products. Thus, the study will use Sentiment Analysis (SA) as the main tool using Python programming language to analyze the customer reviews on dietary supplement products on Amazon.com over the last two years. Moreover, in order to complement an support the analysis performed with Sentiment Analysis (SA), thematic analysis of interviews with fifteen customers of Amazon.com will be performed as part of this studies.As was stated previously, the problem that will be studied in this research will embed how the purchase of dietary supplements products on Amazon.com is directly affected by the positive or negative review the  on specific product.



  1. Relevant (background) literature

Sentiment Analysis (SA) is used to analyze customer reviews, which include various textual elements like words, expressions, and star ratings. Customer review Sentiment Analysis (SA) has the possibility of increasing product sales. Numerous sentiment analysis studies have already been conducted using machine learning techniques. To process customer reviews correctly and accurately, sentiment analysis relies on emotion identification, classification, opinion mining, and sentiment summarization. As part of this study, the literature’s limitations will be analyzed and suggestions of new directions for future scholars to explore will be done. By analyzing customer reviews on Amazon.com, researchers can gain insight into the emotional states of customers, which can then be used to improve product and service offerings. As was mentioned, customers usually have a higher propensity to purchase goods that other consumers have already purchased and found satisfactory. It may be hard to retain the current clients and attract new ones if there are few satisfied customers and skepticism about the validity of consumer reviews is present. Marketplace companies need to develop more compelling and efficient customer interaction methods. Amazon.com controls 77 percent out of 100 percent of the US e-commerce pharmaceutical and supplement sales. Thus, Amazon’s research team determined that by introducing informational pages about the product categories and individual products, more comprehensive product descriptions and improved search results, the shopping experience for customers in this sector would be improved. The quality of the product information and photographs of the product on the listing are equally important to establish an Amazon.com presence. Researchers found that the product descriptions for the Supplement category needed more details, such as dosage instructions, that may have helped customers decide which goods to purchase. Even while using numerous images for products in this category could appear redundant, adding plain words to highlight features or crucial ingredients improves brand messaging and boosts conversion. As a result of the dominance of relationship marketing, customers are increasingly thought to be involved in every stage of production (Wilson et al., 2016).

According to Wassan et al. (2021), the new aspects of the data analysis concepts related to the online and the digital economy have been focused upon. Online Social Networks (OSN), wikis, tagging, folksonomies, blogs, and podcasts are the areas where the majority of social communication from around the world takes place, and that is why SNA, which is known as the social network analysis, is the most critical when it comes to the consumer sentiment analysis nowadays. According to the study, the words, vocabulary and lexicons used in social media help better understand the actual sentiments of the customers. Using machine learning approaches, consumer sentiments can be used to level up sales and client-shared advertising practices. The clients post reviews of amazon products on their social media accounts as well, and the e-commerce companies like Amazon need to make sense of the feedback from the customer or a group of customers buying the same or similar products, which can have a good impact on the sales and the advertising campaigns. The main critical finding of the study is that the two elements of the consumer sentiment analysis are checking for polarity and subjectivity, where the former relates to negative and the positive aspects of the reviews and feelings, and the latter applies to feelings, opinions, sentences, and post content itself. The test was done with about 28,000 ratings and reviews from the e-commerce platform, and the different social media posts of the users were analyzed. The semantics was the critical aspect of the social media analysis.

According to Zhang (2019), the reviews of the Amazon product Alexa have been done, and the data analysis has been done in the Python libraries. The study used Logistic Regression and Naive Bayes, the two machine learning algorithms. This helped discover that the consumer reviews about buying of the Amazon product Alexa were how truly negative or positive an experience. Natural language processing is a critical technology that can be used for the understanding of human words and sentiments and using the machine learning, more sense of the customer’s feelings and the behaviors can be done more effective. According to the study however, machine learning algorithms can be used to understand the texts and the words in a simpler manner and the works and sentences that represent sarcasm, jokes or complex human sentiments are still hard to analyze by the platforms that implicates that more critical development in the machine learning algorithms are needed. However, still, the natural language processing done by the Python platforms can be used to develop a better understanding of the emotions of the customers such as the sadness, happiness as well as the anger, which is critical for better product development and advertisement in the future. The words and the sentences used and written by the customers in the amazon review sections and in the social media websites are critical to be decoded but as they have a cultural and a semantic context, this gives rise to subjectivity and true analysis of it, is kind of not possible using machine learning that are presently available.

According to Hamdallah, A.R., 2021, consumer sentiments should be analyzed and understood effectively to better the experiences associated with the customers’ journey during, before and after the purchase. Moreover, it is implicated in the study that the different kinds of emotions as well as the critical kinds of feelings, are subjective when it comes to different persons and subjective information such as words usage and sentences, and dialect formations are hard to be directly be broken down to emotions and sentiments that are vital to note. The critical finding of the study is that the interaction of the clients and the audiences changes. At the same time, they interact more with the amazon platform, and that is why the change of mood and the change of sentiments can be analyzed and had to be analyzed using the algorithmic and the machine learning techniques that are vital. Ease of shopping, customer service, quality, value, and prices are the different factors that affect the consumer experience of buying or selecting the products. Therefore, the consumer sentiments elicited through the use of these factors are critical to be analyzed. The consumer sentiment analysis and data modelling should be able to make sense of the correlation between the dependent and the independent variables in the consumer experience and consumer journey. Using word clouds, the different kinds of sentiment analysis results can be displayed, which will help in understanding the product and the customer reviews on amazon effectively. Data cleaning and checking for the kappa scores is critical for the right analysis, which should be done.

The star rating of the product on Amazon.com is a common decision-making tool for reviews because it can quickly gauge a reviewer’s mood. Although star ratings are subject to some limitations, they frequently exhibit bias. It is possible for a star rating to be high but the review to be unfavorable, or vice versa. For example, the star rating and review sentiment might not accord in these circumstances. Similar conclusions were reached by Kordzadeh (2019), which found that biases based on the origins of the star ratings’ publication may undermine the reliability of these ratings Therefore, this research suggests that caution must be taken when assessing the reliability of star ratings. Sentiment classification has emerged as a viable solution to fill the gaps left by star ratings and review sentiment. Sentiment classification provides an effective way to capture the general attitude of a piece of text and could be used as a metric for gauging the trustworthiness of reviews. The level of customer satisfaction on Amazon.com is measured by a rating between 1 to 5 stars, where 1 star stands for dissatisfied customer rating and 5 stars mean a good customer review. Additionally, customers can write briefly or extensively a description of why they are satisfied or dissatisfied with the purchased product. Reviews can either harm or benefit the online business. A giant marketplace as Amazon.com is not an exception. Customer reviews affect how consumers see the business and advertise of a product on a marketplace. Analysis of customer feedback in the form of reviews on Amazon with Sentiment Analysis (SA) was studied by different researchers (Nandal, Tanwar, and Pruthi, 2020; Rashid and Huang, 2021; Wassan, Chen and Shen, 2020). However, studies related to consumer reviews on dietary supplements on Amazon.com, as well as related to the analysis of the SA findings are limited (Sharma, 2020). Apart from that, dietary supplements have gained drastic popularity among consumers in the US in recent years (Masterson, 2019).

In the past few years there were different research conducted on Sentiment Analysis (SA) and efforts put on understanding the sentiment of customers’ reviews on Amazon.com. Rashid and Huang (2021) have conducted the Sentiment Analysis (SA) in order to understand the Amazon.com user review dataset. Researchers have conducted the ratings of the products from one to five ratio, where they resulted in the correlation between the product price and the number of helpful reviews and feedback. A similar study was conducted by AlQahtani (2021) where there was a model which would conduct the Sentiment Analysis (SA) on the product reviews on Amazon.com. Other than that, Haque, Saber, and Shah (2018) have conducted a study on Sentiment Analysis (SA) on large-scale Amazon product reviews. Thus, the aim of their study was to categorize the positive and negative feedback of consumers over different products on Amazon.com and develop a supervised learning model to polarize a large amount of review data. However, the study lacks the automation of the processes of product labeling and has caused challenges in the manual implementation of big data. As was mentioned earlier this study will help to give recommendations for further studies on customer experience and particularly customer satisfaction in purchasing dietary supplement products such as vitamins, minerals and herbs on Amazon.com.

Purpose of the study

The purpose of the study is to analyze customer reviews on dietary supplement products on Amazon.com over the last two years. It is vital to study and analyze the components of written customer review of the product to reveal the factors affecting the initial purchasing decision and subsequent buying behavior. By the nature, a person will be more meticulous towards purchasing the product directly related to the health and well-being and indeed will spend time to read the reviews of other customers who have an experience of purchasing and using the supplement. The trend of following the healthy lifestyle mostly influenced by external factors such as celebrities, influencers on social media channels and advertisement have boosted the production and sales of dietary supplement products in the US in the last few years. Undoubtedly, the direct effect of SARS-CoV-19 is unavoidable and have ‘pushed’ individuals to act on their instincts and start purchasing the healthy additives, such as herbs, pills, gummies and others. It is important to relate and further analyze and study the positive and negative reviews of customers on dietary supplement products on Amazon.com to initiating purchasing decision and building a buying behavior. Moreover, the outcomes of the study will help businesses on Amazon.com to reveal and focus on components which have prompted to write a negative or positive review on a product. Thus, the businesses who produce and sell dietary supplement products on Amazon.com will have tools to raise the product sales.