Impact of recommender algorithms on the sales of e-commerce websites

Purpose: With the current pandemic situation, the world is shifting to online buying and therefore the purpose of this study is to understand how the industry can improve sales based on the product recommendations shown on their online platforms. Design/methodology/approach: This paper has studied content-based filtering using decision trees algorithm and collaborative filtering using K-nearest neighbour algorithm and measured their impact on sales of product of different genres on e-commerce websites and if their recommendation causes a difference in sales.This paper has conducted a field experiment to analyse the customer frequency, change in sales caused by different algorithms and also tried analysing the change in buying preferences of customers in post-pandemic situation and how this paper can improve on the search results by incorporating them in the already used algorithms. Findings: This study indicates that different algorithms cause differences in sales and score over each other depending upon the category of the product sold. It also suggests that post-Covid, the buying frequency and the preferences of consumers have changed significantly. Research limitations/implications: The study is limited to existing users of these sites, it also requires the sites to have a huge database of active users and products. Also, the preferences and likings of Indian subcontinent might not generally apply everywhere else. Originality/value: This study enables better insight into consumer behaviour, thus enabling the data scientists to design better algorithms and help the companies improve their product sales.