Interface design for user decision improvement in e-commerce

Customer reviews in E-commerce are playing an important and unique role; a staggering 90 percent of people use and monitor reviews in their online purchasing process. However, the overwhelming number of reviews and inconsistent writing style require significant effort to read and tend to let important information slip by. To help users effectively and efficiently glean information from reviews, a number of systems have summarized customer reviews by extracting features and associate sentiment toward each feature. From the perspective of customers, online purchasing can be viewed as a decision making process. In light of human decision-making theory, we learn that the foundation of effective information displays for user decision improvement is gaining a deep understanding of user decision-making behavior. However, no clear picture exists to systematically elaborate on consumer decision-making behavior in E-commerce, in particular, with respect to customer reviews. In this thesis, we take online hotel booking as an example to empirically investigate consumer decision-making behavior in three stages of online purchasing: (1) screening out interesting alternative(s) for further consideration, (2) evaluating alternatives in detail, and (3) comparing candidates to make the final choice. Interfaces that aggregate information from customer reviews have been developed to support the three alternative stages. Through analysis of the results, we identify the decision strategies users utilize to process information and the information they are inclined to seek at each stage. These findings lay solid groundwork to design E-commerce interfaces for consumer decision improvement. Concerning user decision-making behavior in the stage of screening out interesting alternatives, we find that: (1) 94% of participants began by eliminating alternatives with values for an attribute below a cut-off to simplify the complexity of the choice; (2) 55.3% of participants eliminated alternatives by both static features (i.e., product specifications) and customer reviews. Moreover, the number of users who adopted opinion attributes (i.e., attributes extracted from customer reviews) is significantly higher than that using an overall review score; and (3) the cut-off values are determined by the value distribution of an attribute and correlation among attributes, in addition to stable preference. Grounded in these user decision-making behaviors, we framed two alternative designs for an opinion-attribute-embedded filter panel based on checkboxes and sliders. In the checkbox interface, the filter for each attribute is represented in the form of an array of N checkboxes, which is utilized in most E-commerce websites. In the slider interface, the filter for each numerical attribute (i.e., price and opinion attributes) is represented by a modified slider, which visualizes the distribution of an attribute via bars and the correlation among attributes via simultaneous change. Then, we performed a user study to compare the two alternative designs in the context of online hotel booking. The results show that people depended highly on opinion attributes to narrow down the range of options in both interfaces, which points to the effectiveness of incorporating opinion attributes in filters. And the slider interface achieves significantly higher user assessments in terms of perceived decision accuracy, cognitive effort, pleasantness to use and intention to return. After narrowing down options to a smaller set, 40% of participants adopted a more compensatory strategy – Weighted Additive Difference, i.e., comparing the remaining alternatives on multiple attributes and selecting the alternative with the best overall value. More notably, significantly more participants compared alternatives by opinion attributes in comparison with those associated with an overall review score. Therefore, we developed a multi-attribute sorting panel embedded with opinion attributes. Furthermore, the multi-attribute sorting panel was expanded to three alternative designs that mainly differ in the way of eliciting relative importance for attributes: (1) direct assessment, asking users to directly assign weights to attributes; (2) indifference method, modifying one of two sets of stimuli until subjects feel that there is no difference between the two; and (3) indirect measurement, giving relative preference on a pair of alternatives. Through analysis of objective and subjective measures, the multi-attribute sorting was verified to be beneficial to consumer online purchasing. The direct way outperforms the indifference and indirect ways regarding perceived decision accuracy, cognitive effort, satisfaction and intent to use in an E-commerce environment. Motivated by the results of the above user studies, we have derived a set of guidelines on how to design interfaces for consumer purchase decision improvement in E-commerce.