Baruch College, The City University of New York, USA
This chapter reviews the different types of personalization systems commonly employed by Web sites and argues that their deployment as Web site interface design decisions may have as big an impact as the personalization systems themselves. To accomplish this, this chapter makes a case for treating Human-Computer Interaction (HCI) issues seriously. It also argues that Web site interface design decisions made by organizations, such as the type and level of personalization employed by a Web site, have a direct impact on the communication capability of that Web site. This chapter also explores the impact of the deployment of personalization systems on users’ loyalty towards the Web site, thus underscoring the practical relevance of these design decisions.
Organizations, in anticipation of the emergence of Web interface as a major point of contact between companies and customers, are beginning to employ a wide variety of technologies to build meaningful relationships with their customers. While Web interface may not be the only point of contact that customers use, organizations are aware of the advantages of using the Web to cater to the needs of the customers. This concept of “self-service” not only reduces costs for the company in the long run, but also increases customer satisfaction by addressing the transactional and the relational needs of the customer. In this chapter, I review the different types of personalization systems commonly employed by Web sites and argue that their deployment as Web site interface design decisions may have as big an impact as the personalization systems themselves. To accomplish this, I make a case for treating HCI issues seriously and argue that Web site interface design decisions made by organizations, such as the type and level of personalization employed by a Web site, has a direct impact on the communication capability of that Web site. I also focus on the practical relevance of these Web site design decisions by examining their effect on users’ loyalty towards the Web site.
Most of the technologies and tools that companies use to manage their relationship with their customers usually fall under the banner of Customer Relationship Management (CRM) System. Even though personalization is just one piece of the CRM pie, it is a very crucial piece as effective personalization significantly enhances the ability of the organization to initiate a discourse with its customers to the point where any and all of these dialogues are seamlessly integrated with the database’s historical and transactional information. Based on the data stored in these databases and recent history (the pages customers viewed in the last session), Web sites automatically attempt to improve their organization and presentation of content. These Web sites, armed with a host of appropriate tools — including intelligent agents, recommendation engines and the like — attempt to anticipate the context of the interaction with their customers and personalize each customer’s shopping experience (Andre & Rist, 2002; Billsus, Brunk, Evans, Gladish, & Pazzani, 2002).
Personalization is a process of providing special treatment to a repeat visitor to a Web site by providing relevant information and services based on the visitor’s interests and the context of the interaction (Chiu, 2000; Cingil, Dogac, & Azgin, 2000). Personalization is needed to successfully manage customer relationships, promote the right product the customer is interested in, and manage content. Most of the advanced personalization might require sophisticated data mining techniques and the ability to display dynamic content without seriously compromising system resources (dynamic display of content will usually mean increased download time).
There are a few well-known techniques for personalization. Rules-based personalization modifies the content of a page based on specific set of business rules. Cross-selling is a classic example of this type of personalization. The key limitation of this technique is that these rules must be specified in advance. Personalization that uses simple filtering techniques determines the content that would be displayed based on predefined groups or classes of visitors and is very similar to personalization based on rules-based techniques. Personalization based on content-based filtering analyzes the “contents of the objects to form a representation of the visitor’s interest” (Chiu, 2000). This would work well for products with a set of key attributes. For example, a Web site can identify the key attributes of movies (VHS, DVD) such as drama, humor, violence, etc., and can recommend movies to its visitors based on similar content. Personalization based on collaborative filtering offers recommendations to a user based on the preferences of like-minded peers. To determine the set of users who have similar tastes, this method collects users’ opinion on a set of products using either explicit or implicit ratings (Chiu, 2000). Please see Figure 8-1 for an illustration of how a Web site could use all three personalization methods to best serve the customer.
Figure 8-1: Overview of personalization techniques
An intelligent way to make the Web site adaptive is to use not only the information provided by the user (such as rating the music and log-in information), but also information that could be collected based on the click-stream trail left behind by the user. These two different sources of collecting information about the consumer are known as explicit and implicit profiling. As the name implies, explicit profiling collects information about a user by directly asking him or her information about himself or herself and product likes and dislikes. This information is collected over a period of time and is stored in the customer database as a profile. Typically, the user would need to log-in in order for the Web site to access the profile and provide personalized content. Even though cookies can be used to store this information on a user’s hard disk, companies prefer to use the log-in approach as this allows the Web site to identify the unique visitor (cookies won’t help if the computer is shared within a family or if the customer accesses the Web site from a different location — say from the office).
Implicit profiling typically tracks the actual behavior of the customer while browsing the Web site. This method of collecting information is transparent to the user. While less intrusive, this method of collecting information has implications for the user’s privacy. Typically, information is collected about the pages the consumer visited, the products he or she looked at and the time that the user spent on these pages. If a (brick and mortar) company has good information systems, the data from explicit and implicit profiling can be merged with the off-line customer information (see legacy user data in Figure 8-1) to effectively present a seamless Web interface to the customer.
Ideally, a company should use all sources of information it has about the customer. However, when a user visits a shopping Web site (even a repeat user), it would be unsound business practice to expect the user to log-in every time to access personalized content. Hence, a good Web site would use implicit profiling and make a few assumptions about the likes and dislikes of the customer to provide adaptive content to the customer. For example, if a customer visits a specific product page, it is a good idea to assume that the customer is interested in that particular product and provide content personalized to that user’s need. Of course, in most cases, even if the user logs in, the Web site may have little else other than previous purchase history if the user has not provided any specific information on the products he or she likes.
The level and extent of personalization offered by the Web site will have an effect on the communication characteristics of the media. This research argues that different levels of support provided for personalization will specifically impact on the adaptiveness [similar to contingency used by (Burgoon et al., 2000)] of the Web site. This is best illustrated by discussing a real life example using Amazon.com. Appendices 1 to 3 include three screen shots that show the different ways Amazon.com attempts to personalize the experience of the customer. When the user enters the Web site, he or she is invited to log in if desired. Once the user logs in, Appendix 1 shows the Web page that is dynamically created by Amazon.com. This page recommends products to the user based on past purchase history and on the explicit ratings provided by the user to a set of select items. Appendix 2 shows the product page for a book the user is interested in. The column on the left hand side of this page shows the associated related content about the product that is displayed on this page. Appendix 3 shows the page tailor-made for the user based on his recent browsing history and past purchase history. Of course, the scenario described above assumes that the user logged into the Web site at the outset. An intelligent Web site can still adapt its content in its product page by assuming that the user is interested in the product he or she is browsing. Accordingly, the product page shown in screen shot 2 can be personalized even without an explicit log-in by the user.
If the same user were to shop for the book that he is interested in a physical store, he might have approached the sales clerk (or even a friend he had taken along for the shopping trip) for help locating the product. Now, when he mentions to his friend that he is interested in this specific book, music or movie, then it is possible to imagine a conversation happening along the lines discussed above. Of course, the above discourse with the Web site is limited by the need for a shared context. The conversation will not be totally indeterminable in terms of context and content and may not move along in any arbitrary direction as is possible in a conversation with a friend. But, this research argues that there are enough cues in the discourse initiated by the personalization system of Amazon.com that is enough to give the user the impression that the conversation is contingent within that shared context.
To enhance the relationship with the customers, companies can also provide support for virtual communities, as this will facilitate access to free-flowing and unstructured information beyond what is provided by the computer agents (Jones, 1997; Preece, 2001, 2002). For example, companies can aggregate the opinions of consumers on a particular product and present them to a new user who is browsing that product page. Depending on the level of support provided by the Web site, the new user can also get in touch with another consumer he or she might identify with, as is the case with Amazon.com. A recent study (Brown, Tilton, & Woodside, 2002) shows that community features create value for a shopping Web site. Their study showed that community users accounted for about one-third of the visitors to the e-tailing sites surveyed and that they also generated twothirds of the sales (2000 transactions worth one million dollars). Practioners have long argued that having a vibrant community in the form consumer reviews is crucial for the success of e-commerce Web sites such as Amazon.com and Ebay.com (Brown et al., 2002; Kirkpatrick, 2002). Hence providing support for consumer reviews facilitates formation of one type of virtual community and integrating high level of support (user rating and information about the user) for consumer reviews on the product page increases personalization afforded by Web sites as these are relevant comments and opinions by different users presented on the product page.
Reeves, Nass and their colleagues at the Center for the Study of Language and Information at Stanford have shown that even experienced users tend to respond to computers as social entities (Nass, Lombard, Henriksen, & Steur, 1995; Nass, Moon, Fogg, Reeves, & Dryer, 1995; Nass & Steur, 1994). These studies indicate that computer users follow social rules concerning gender stereotypes and politeness, and that these social responses are to the computer as a social entity and not to the programmer. When explicitly asked by the researchers, most users consistently said that social responses to computers were illogical and inappropriate. Yet, under appropriate manipulation, they responded to the computer as though it were a social entity. This, in fact, is the essence of the Theory of Social Response (Moon, 2000; Reeves et al., 1997). Thus I argue that there is value in conceptualizing the Web site as a social actor and that the Web site can be equated to the “agents” mentioned above in terms of source orientation. There are several points-of-contact between a Web site and its users that will result in responses by the users not unlike the way they would respond to a social interaction.
In the light of the above discussions, Web sites should also view deployment of personalization systems as important Web site design decisions that will facilitate or hinder this interactive dialogue between a Web site and its users. In a recent study conducted by this author, the personalization systems deployed by four major Web sites (Amazon.com, BarnesandNoble.com, CDNow.com and Chapters.ca) were compared along with the Web sites’ support for virtual communities. The results of the study showed strong support showing that level of support for personalization systems had an impact on customer loyalty. The results also suggested that Web sites should pay close attention to the way they deploy these personalization systems. Specifically, the results showed that by implementing consumer review support on the product page (as done by Amazon.com), we could simulate support for personalization systems in the absence of personalization systems.
In practice, it is advantageous for the Web sites to offer some form of support for personalization or virtual community as this makes the Web site to be perceived as more adaptive. This will facilitate better communication between the Web site and the shoppers, thus leading to higher levels of customer loyalty. Companies do understand that in practical terms it takes a lot more money and effort to acquire a new customer than to keep an existing customer and arguments presented in this chapter throws new light on the role of support for personalization and consumer reviews in increasing customer loyalty.
Good personalization systems can be very expensive to set up. Enabling an e-commerce Web site with the necessary tools to build a vibrant community costs little (especially when compared to personalization systems), as the community members provide the content. The results of the study offer evidence that Web sites by providing support for consumer reviews could reap the benefits in terms of increased adaptiveness despite offering very little personalization. However, the ways and means of implementing support for personalization and virtual communities deserve further research. Future research should examine the implementation of these features in finer detail. This will help the organizations understand more in depth the trade-offs involved in providing different types and levels of personalization and virtual communities.
Andre, E., & Rist, T. (2002). From adaptive hypertext to personalized Web companions. Communications of the ACM, 45(5), 43-46.
Billsus, D., Brunk, C. A., Evans, C., Gladish, B., & Pazzani, M. (2002). Adaptive interfaces for ubiquitous Web access. Communications of the ACM, 45(5), 34-38.
Brown, S., Tilton, A., & Woodside, D. (2002). Online communities pay. McKinsey Quarterly, 19(1), 17.
Burgoon, J. K., Bonito, J. A., Bengtsson, B., Cederberg, C., Lundeberg, M., & Allspach, L. (2000). Testing the interactivity model: Communication processes, partner assessments, and the quality of collaborative work. Journal of Management Information Systems, 16(3), 33-56.
Chiu, W. (2000). Web site personalization. IBM High-Volume Web Site Team, WebSphere Software Platform.
Cingil, I., Dogac, A., & Azgin, A. (2000). A broader approach to personalization. Communications of the ACM, 43(8), 136-141.
Jones, Q. (1997). Virtual-communities, virtual settlements & cyber-archaeology: A theoretical outline. Journal of Computer Mediated Communication, 3(3).
Kirkpatrick, D. (2002). Esther Dyson: Living the networked life. Fortune, 145(11), 168-172.
Moon, Y. (2000). Intimate exchanges: Using computers to elicit self-disclosure from consumers. Journal of Consumer Research, 26, 323-339.
Nass, C., Lombard, M., Henriksen, L., & Steur, J. (1995). Anthropocentrism and computers. Behavior and Information Technology, 14(4), 229-238.
Nass, C., Moon, Y., Fogg, B. J., Reeves, B., & Dryer, C. (1995). Can computers be human personalities? International Journal of Human-Computer Studies, 43(2), 223-239.
Nass, C., & Steur, J. (1994). Voices, boxes and sources of messages: Computers and social actors. Human Communication Research, 19(4), 504-527.
Preece, J. (2001). Sociability and usability: Twenty years of chatting online. Behavior and Information Technology, 20(5), 347-356.
Preece, J. (2002). Supporting community and building social capital. Communications of the ACM, 45(4), 37-39.
Reeves, B., & et al. (1997). The media equation. Stanford, CA: CSLI Publications.
Section I - Consumer Behavior in Web-Based Commerce
Section II - Web Site Usability and Interface Design
Section III - Systems Design for Electronic Commerce
Section IV - Customer Trust and Loyalty Online
Section V - Social and Legal Influences on Web Marketing and Online Consumers