Today's digital marketplaces offer consumers great convenience, immense product choice, and large amounts of product-related information. However, as a result of the cognitive constraints of human information processing, finding products that satisfy consumers' needs and/or interests is not an easy task. Many online stores have implemented technological tools to assist consumers in product search and selection. One such tool is web-based product recommendation agents (PRAs), a type of web personalization technology that provides individual consumers with product recommendations based on their product-related needs and preferences expressed [[1], [2], [3]]. PRAs have the potential to help consumers make effective purchase decisions [4,5] and reduce consumers' uncertainty about buying products online [6].
Personalized recommendations are an e-commerce feature that is highly valued by consumers and also help online companies create brand loyalty [7]. This explains the phenomenal increase in the adoption of PRAs by online merchants in various industries (e.g., Amazon, eBay, Apple) [8]. Despite the potential for PRAs to aid consumers in decision making, consumers are empowered to the extent that the PRAs provide true personalization by recommending products based solely on and thus best represent consumers' preferences [9,10]. Just as a salesperson may simultaneously serve two principals (i.e., the client and the employing company) [11,12], a PRA may also be designed to serve the interest of its users (i.e., consumers) as well as that of its provider, creating opportunities for competing loyalties [11]. The objectives and economic incentives of the PRA providers (e.g., online merchants) are not always necessarily aligned with those of the consumers [13]. To balance their competing goals of aiding consumers in decision making and generating more business opportunities [8,14], online merchants may implement PRAs to provide recommendations that are not solely preference-matched to benefit consumers but instead are biased toward their own interests [3,8,15]. This may result in recommendations for products from certain vendors and/or with certain characteristics (e.g., high profit-margin products, soon-to-be discontinued products) so as to attain higher-than-usual profits (in terms of mark-ups, commissions, referral fees, etc.) [16,17] or reduce losses.
There have already been reported incidents of online companies (e.g., Amazon, Hotel.de) using PRAs to provide biased recommendations to consumers [14,18,19]. Prior academic research has also observed that PRAs have the potential to not only assist consumers but also steer them in a particular direction (e.g., [[20], [21], [22], [23], [24]]). However, to our knowledge, there has been limited understanding of the underlying mechanisms by which biased recommendations influence consumers' decision making outcomes in online buying, which is a gap the current study aims to fill.
Drawing on the theoretical perspective of personalization, this study examines how biased personalized product recommendations from PRAs influence consumers' decision quality and decision effort when shopping online. It further explores the role perceived personalization plays in mediating the impact of biased PRAs on consumer decision making. Specifically, this study compares the relative impact of biased personalized recommendations with that of biased non-personalized recommendations (based on false claim of popularity) on consumers' decision quality and decision effort. We posit that the availability of a PRA on a website may enhance a consumer's belief that due to personalization (be it true or not) her product search would be more effective and efficient. If the biased personalized recommendations do not result in improved objective decision quality for the consumer, while at the same time these recommendations lead the consumer to believe that her product choice is an optimal one, leading to her continued reliance on the PRA, then unscrupulous online merchants will have a powerful means with which they can influence consumer decision making toward the merchants' own benefits. With its focus on undesirable consequences of PRA use, this study answers the call for more academic research into the dark side of information technology (IT) [25].
The remainder of this paper is organized as follows. Section 2 presents the study's theoretical foundation and develops the hypotheses. The research method and results of hypothesis testing are reported in 3 Research method, 4 Data analysis and results. And the paper concludes with a discussion of the results, limitations, and contributions of the study and suggests areas for future research.