Abstract:
The purpose of this study is to determine whether there exists an impact of personalization in retail websites with the implementation of collaborative filtering and content-based filtering recommendation techniques on consumer satisfaction. The main objective is to understand the possible impact of personalization on product rating as an equivalent measure of consumer satisfaction and to compare the two recommendation techniques. Three distinct statistical approaches, including Analysis of Variance, linear regression, and logistic regression, are used to analyze the data collected from Amazon using a web scraping tool, Mozenda. The analysis conducted looks at the differences between the mean product ratings of recommended and non-recommended products, and according to the results, the products displayed under the Collaborative Filtering and Content-Based Filtering recommendation sections are rated higher, leading to higher satisfaction among consumers. However, contrary to the previous studies, when comparing both techniques together, no difference is observed in their impact on consumer satisfaction. In addition, the study also explores the impact of other product information such as
recommendation rates and written reviews analyzed through sentiment analysis on the satisfaction of consumers but also on the likelihood of one technique being chosen to recommend a product over the other. While the study provides valuable insights, it also addresses the limitations, especially in the availability of primary and internal data, which could be included in future research. The findings could be useful to Amazon’s business analysts, the web scraping company, and future researchers by providing a deeper understanding of the impact of personalization of retail websites in enhancing consumer satisfaction.
